CTNF 18/361,238 CTNF 101543 DETAILED ACTION Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Information Disclosure Statement 06-52 The information disclosure statement (IDS) submitted on October 23, 2023. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. Claim Objections 07-29-01 AIA Claim 13, line 2 is objected to because of the following informalities: The claim recites “ The method of claim 11, further comprising repeating the obtaining, inputting, inputting, calculating ,…”, where there is a repeated word “ inputting ” . Appropriate correction is required. Claim Rejections - 35 USC § 101 07-04-01 AIA 07-04 35 U.S.C. 101 reads as follows: Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title. Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea (mental process or math concept) without significantly more. Claim 1: Regarding claim 1, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “A method, comprising: inputting first training ultrasound data, including a fundamental component and a harmonic component, to each of a plurality of teacher models, and training each teacher model of the plurality of teacher models with the first training ultrasound as teacher input data and second training ultrasound data, including the harmonic component, as teacher target data; acquiring, for each teacher model of the plurality of the trained teacher models, corresponding first estimated data output from the teacher model, in response to input of first ultrasound data to the teacher model; selecting a first particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models; and training a student model with the first ultrasound data as student input data and the corresponding first estimated data of the selected first particular teacher model as student target data ,” and a method is one of the four statutory categories of invention. In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components: selecting a first particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models; (This is a mental process, a person can mentally evaluate and select a teacher model by evaluating corresponding data output from the trained teacher models, see MPEP 2106.04(a)(2)(III)), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application: A method, comprising: inputting first training ultrasound data, including a fundamental component and a harmonic component, to each of a plurality of teacher models, ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g)), and training each teacher model of the plurality of teacher models with the first training ultrasound as teacher input data and second training ultrasound data, including the harmonic component, as teacher target data; ( This recites mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), acquiring, for each teacher model of the plurality of the trained teacher models, corresponding first estimated data output from the teacher model, in response to input of first ultrasound data to the teacher model; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g)), and training a student model with the first ultrasound data as student input data and the corresponding first estimated data of the selected first particular teacher model as student target data, ( This recites mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea. In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, additional element iii and v recite mere instructions to apply the judicial exception using generic computer components, which are not indicative of significantly more. The additional elements ii and iv recite mere data gathering, and are considered insignificant extra-solution activities. In step 2B, these insignificant extra-solution activities are well understood routine and conventional activities which include receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)), Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea . Therefore, the claim is not patent eligible. Claim 2: Regarding claim 2, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 2 recites the following additional limitation: The method of claim 1, wherein each of the plurality of trained teacher models has a different model structure and more parameters than the student model , (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 3: Regarding claim 3, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 3 recites the following abstract idea: selecting a second particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding second estimated data output from each of the trained teacher models; ( This is considered a mental process, since a person can mentally evaluate and select a teacher model from evaluating corresponding data output from the trained teacher models, see MPEP 2106.04(a)(2)(III)), Further, claim 3 recites the following additional elements: The method of claim 1, further comprising: acquiring, for each teacher model of the plurality of the trained teacher models, corresponding second estimated data output from the teacher model, in response to input of second ultrasound data, different from the first ultrasound data, to the teacher model; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) – see MPEP 2106.05(d) (II)(i), and training the student model with the second ultrasound data as the student input data and the corresponding second estimated data of the selected second particular teacher model as the student target data , (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 4: Regarding claim 4, it is dependent upon claim 3, and thereby incorporates the limitations of, and corresponding analysis applied to claim 3. Further, claim 4 recites the following additional element: The method of claim 3, wherein the selected second particular teacher model is different from the selected first particular teacher model, (In step 2A, prong 2, this recites an indication to a field of use or technological environment – see MPEP 2106.05(h)), (In step 2B, this also recites a field of use or technological environment – see MPEP 2106.05(h)), Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 5: Regarding claim 5, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 5 recites the following additional element: The method of claim 1, further comprising: generating output ultrasound data, including the harmonic component, by inputting input ultrasound data into the trained student model , ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) – see MPEP 2106.05(d) (II)(i), Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 6: Regarding claim 6, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “ An apparatus, comprising: processing circuitry configured to input first training ultrasound data, including a fundamental component and a harmonic component, to each of a plurality of teacher models, and train each teacher model of the plurality of teacher models with the first training ultrasound as teacher input data and second training ultrasound data, including the harmonic component, as teacher target data; acquire, for each teacher model of the plurality of the trained teacher models, corresponding first estimated data output from the teacher model, in response to input of first ultrasound data to the teacher model; select a first particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models; and train a student model with the first ultrasound data as student input data and the corresponding first estimated data of the selected first particular teacher model as student target data,” and an apparatus (or machine) is one of the four statutory categories of invention. In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components: select a first particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models; ( This recites a mental process , a person can mentally evaluate and select a teacher model from evaluating data outputs of trained teacher models, see MPEP 2106.04(a)(2)(III)), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application: An apparatus, comprising: processing circuitry configured … ( This recites mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), to input first training ultrasound data, including a fundamental component and a harmonic component, to each of a plurality of teacher models ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g)), and train each teacher model of the plurality of teacher models with the first training ultrasound as teacher input data and second training ultrasound data, including the harmonic component, as teacher target data; ( This recites mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), acquire, for each teacher model of the plurality of the trained teacher models, corresponding first estimated data output from the teacher model, in response to input of first ultrasound data to the teacher model; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g)), and train a student model with the first ultrasound data as student input data and the corresponding first estimated data of the selected first particular teacher model as student target data , ( This recites mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea. In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, additional element ii, iv, and vi recite mere instructions to apply the judicial exception using generic computer components, which are not indicative of significantly more. The additional elements iii and v recite mere data gathering, and are considered insignificant extra-solution activities. In step 2B, these insignificant extra-solution activities are well understood routine and conventional activities which include receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)), Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea . Therefore, the claim is not patent eligible. Claims 7-10: Regarding claim 6, all of claim 6’s dependent claims follow the deficiencies of their parent claim. Since claims 7-10 recite similar limitations as corresponding claims 2-5 listed above, and are rejected for similar reasons under 35 U.S.C. 101. Claim 11: Regarding claim 11, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “ A method, comprising: obtaining first ultrasound data, including a fundamental component and a harmonic component, as input ultrasound data and second ultrasound data, including the harmonic component, as target output ultrasound data corresponding to the first ultrasound data; inputting the first ultrasound data to a previously trained teacher model to generate teacher output ultrasound data; inputting the first ultrasound data to a student model to generate student output ultrasound data; calculating a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data; and updating parameters of the student model based on the calculated loss value ” , and a method is one of the four statutory categories of invention. In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a math concept but for recitation of generic computer components: calculating a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data; ( This recites a mathematical relationship, mathematical formula or equation, or mathematical calculation , see pages 12-13 from the specification state “In one embodiment, the online training process can be represented by the following pseudo code, in which (F s ,θ s ) represents the student model 474 , where θ s is the set of trainable parameters; (F T , θ T ) represents the pseudo-teacher model 482 , where θ T is the set of frozen parameters; L(Z S , y′; θs) is the loss between ground truth y′ (desired IQ output 464 ) and the student model output Z S ( 468 ) for model input x′; L(Z S , Z T ; θ S ) is the loss between the pseudo best teacher model output Z T ( 484 ) and the student model output Z S ( 468 ); and λ is loss weight.” PNG media_image1.png 586 726 media_image1.png Greyscale , see MPEP 2106.04(a)(2), subsection I), and updating parameters of the student model based on the calculated loss value , ( This recites a mathematical relationship, mathematical formula or equation, or mathematical calculation , see pages 12-13 from the specification state “In step S910, the method includes updating parameters of the student model based on the calculated loss value”, see MPEP 2106.04(a)(2), subsection I), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mathematical concept but for the recitation of generic computer components, then it falls within the mathematical concept grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application: A method, comprising: obtaining first ultrasound data, including a fundamental component and a harmonic component, as input ultrasound data and second ultrasound data, including the harmonic component, as target output ultrasound data corresponding to the first ultrasound data; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activit y – see MPEP 2106.05(g)), inputting the first ultrasound data to a previously trained teacher model to generate teacher output ultrasound data; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activit y – see MPEP 2106.05(g)), inputting the first ultrasound data to a student model to generate student output ultrasound data; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activit y – see MPEP 2106.05(g)), Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea. In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional elements iii, iv, and v recite mere data gathering, and are considered insignificant extra-solution activities. In step 2B, these insignificant extra-solution activities are well understood routine and conventional activities which include receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)), Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea . Therefore, the claim is not patent eligible. Claim 12: Regarding claim 12, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 12 recites the following abstract ideas: The method of claim 11, wherein the calculating step further comprises: calculating a first loss value of a first loss function based on the generated student output ultrasound data and the target output ultrasound data; This is considered a mathematical relationship, mathematical formula or equation, or mathematical calculation , see specification in pages 12-13 stating “In one embodiment, the online training process can be represented by the following pseudo code, in which (F s ,θ s ) represents the student model 474 , where θ s is the set of trainable parameters; (F T , θ T ) represents the pseudo-teacher model 482 , where θ T is the set of frozen parameters; L(Z S , y′; θs) is the loss between ground truth y′ (desired IQ output 464 ) and the student model output Z S ( 468 ) for model input x′; L(Z S , Z T ; θ S ) is the loss between the pseudo best teacher model output Z T ( 484 ) and the student model output Z S ( 468 ); and λ is loss weight.” Specification notes that loss values are calculated from the algorithm listed on page 13, PNG media_image1.png 586 726 media_image1.png Greyscale see MPEP 2106.04(a)(2), subsection I), calculating a second loss value of a second loss function based on the generated teacher output ultrasound data and generated student output ultrasound data; This is considered a mathematical relationship, mathematical formula or equation, or mathematical calculation , see specification in pages 12-13 similar to above, see MPEP 2106.04(a)(2), subsection I), and calculating the loss value as a weighted sum of the first loss value and the second loss value , (This is considered a mathematical relationship, mathematical formula or equation, or mathematical calculation , see specification in pages 12-13 similar to above, see MPEP 2106.04(a)(2), subsection I), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mathematical concept but for the recitation of generic computer components, then it falls within the mathematical concept grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 13: Regarding claim 13, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 13 recites the following abstract idea: … calculating … until the parameters of the student model satisfy a convergence criteria , ( This recites a mathematical relationship, mathematical formula or equation, or mathematical calculation, see from specification on pages 12-13 states “In one embodiment, the online training process can be represented by the following pseudo code, in which (F s ,θ s ) represents the student model 474 , where θ s is the set of trainable parameters; (F T , θ T ) represents the pseudo-teacher model 482 , where θ T is the set of frozen parameters; L(Z S , y′; θs) is the loss between ground truth y′ (desired IQ output 464 ) and the student model output Z S ( 468 ) for model input x′; L(Z S , Z T ; θ S ) is the loss between the pseudo best teacher model output Z T ( 484 ) and the student model output Z S ( 468 ); and λ is loss weight.” PNG media_image1.png 586 726 media_image1.png Greyscale The convergence criteria are part of a math calculation, see MPEP 2106.04(a)(2), subsection I), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mathematical concept but for the recitation of generic computer components, then it falls within the mathematical concept grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. Further, claim 13 recites the following additional element: The method of claim 11, further comprising repeating the obtaining, inputting, inputting, …, and updating steps for different input ultrasound data and corresponding different second ultrasound data , ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) – see MPEP 2106.05(d) (II)(i), Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 14: Regarding claim 14, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 14 recites the following abstract idea: The method of claim 11, wherein the updating step comprises updating the parameters of the student model without updating parameters of the trained teacher model , ( This is considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see pages 12-13 of an algorithm for training and updating parameters and page 13 state “in step S 910 , the method includes updating parameters of the student model based on the calculated loss value”, and page 14 from the specification state "FIG. 11 is a flowchart of the updating step of FIG. 9 . In step S1102, the method includes updating the parameters of the student model without updating parameters of the trained teacher model". Since the updating step involves at least one calculated loss value, this recites a math calculation, see MPEP 2106.04(a)(2), subsection I), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mathematical concept but for the recitation of generic computer components, then it falls within the mathematical concept grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 15: Regarding claim 15, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 15 recites the following additional element: The method of claim 11, further comprising: generating output ultrasound data, including the harmonic component, by inputting input ultrasound data into the trained student model, ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) – see MPEP 2106.05(d) (II)(i), Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible. Claim 16: Regarding claim 16, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “ An apparatus, comprising: processing circuitry configured to obtain first ultrasound data, including a fundamental component and a harmonic component, as input ultrasound data and second ultrasound data, including the harmonic component, as target output ultrasound data corresponding to the first ultrasound data; input the first ultrasound data to a previously trained teacher model to generate teacher output ultrasound data; input the first ultrasound data to a student model to generate student output ultrasound data; calculate a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data; and update parameters of the student model based on the calculated loss value” , and an apparatus or machine is one of the four statutory categories of invention. In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a math concept but for recitation of generic computer components: calculate a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data; ( This recites a mathematical relationship, mathematical formula or equation, or mathematical calculation , see pages 12-13 from the specification state “In one embodiment, the online training process can be represented by the following pseudo code, in which (F s ,θ s ) represents the student model 474 , where θ s is the set of trainable parameters; (F T , θ T ) represents the pseudo-teacher model 482 , where θ T is the set of frozen parameters; L(Z S , y′; θs) is the loss between ground truth y′ (desired IQ output 464 ) and the student model output Z S ( 468 ) for model input x′; L(Z S , Z T ; θ S ) is the loss between the pseudo best teacher model output Z T ( 484 ) and the student model output Z S ( 468 ); and λ is loss weight.” PNG media_image1.png 586 726 media_image1.png Greyscale , see MPEP 2106.04(a)(2), subsection I), and update parameters of the student model based on the calculated loss value , ( This recites a mathematical relationship, mathematical formula or equation, or mathematical calculation , see pages 12-13 from the specification state “In step S910, the method includes updating parameters of the student model based on the calculated loss value”, see MPEP 2106.04(a)(2), subsection I), If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mathematical concept but for the recitation of generic computer components, then it falls within the mathematical concept grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea. In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application: An apparatus, comprising: processing circuitry configured… ( This recites mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), …to obtain first ultrasound data, including a fundamental component and a harmonic component, as input ultrasound data and second ultrasound data, including the harmonic component, as target output ultrasound data corresponding to the first ultrasound data; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activit y – see MPEP 2106.05(g)), input the first ultrasound data to a previously trained teacher model to generate teacher output ultrasound data; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activit y – see MPEP 2106.05(g)), input the first ultrasound data to a student model to generate student output ultrasound data; ( In step 2A, prong 2, this recites mere data gathering, which is considered insignificant extra-solution activit y – see MPEP 2106.05(g)), Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea. In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, additional element iii recites mere instructions to apply the judicial exception using generic computer components, which is not indicative of significantly more. The additional elements iv, v, and vi recite mere data gathering, and are considered insignificant extra-solution activities. In step 2B, these insignificant extra-solution activities are well understood routine and conventional activities which include receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)), Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea . Therefore, the claim is not patent eligible. Claims 17-20: Regarding claim 16, all of claim 16’s dependent claims follow the deficiencies of their parent claim. Since claims 17-20 recite similar limitations as corresponding claims 12-15 listed above, and are rejected for similar reasons under 35 U.S.C. 101. Claim Rejections - 35 USC § 103 07-06 AIA 15-10-15 In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status. 07-20-aia AIA The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action: A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made. 07-23-aia AIA The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows: 1. Determining the scope and contents of the prior art. 2. Ascertaining the differences between the prior art and the claims at issue. 3. Resolving the level of ordinary skill in the pertinent art. 4. Considering objective evidence present in the application indicating obviousness or nonobviousness. 07-21-aia AIA Claim s 1, 2, 6 and 7 are rejected under 35 U.S.C. 103 as being unpatentable over Rubin, J. et al., in “Attention distillation for detection transformers: application to real-time video object detection in ultrasound,” published on December 1, 2021, available at https://proceedings.mlr.press/v158/rubin21a/rubin21a.pdf, (hereafter, RUBIN), in view of Li, X. et al., (Pub. No. CN115861132A), published on March 28, 2023, (hereafter, LI), further in view of Zhang, Y. et al., (Pub. No. CN113487614B), published on November 30, 2021, (hereafter, ZHANG) . Claim 1: Regarding claim 1, RUBIN teaches “a method , comprising: … inputting first training ultrasound data , … to each of a plurality of teacher models, and training each teacher model of the plurality of teacher models with the first training ultrasound as teacher input data and second training ultrasound data, including the harmonic component, as teacher target data;” See RUBIN in page 33, section 5.3. 3D-to-2D Attention Distillation for Videos, describes "Here, we utilized a 3D-DETR R50-6/6 model as the teacher network. The network applies a series of k=6 spatio-temporal 3D convolutions to a stack of ultrasound video frames. The effect is that the temporal dimension is convolved out to arrive at a 2D feature map with a singular temporal dimension, k=1, while image width and height remains unchanged. The learned feature map is then fed to the remainder of the network that has the same architecture as the 2D DETR-R50-6/6 teacher network." Here, RUBIN describes generating teacher models for both 2D and 3D ultrasound images, indicating a plurality of teacher models. Additionally, RUBIN shows that for each teacher model, a corresponding feature map (i.e. image output ) was generated for that model. See Tables 1 and 2, and figure 2 which inputs ultrasound frames. RUBIN describes using the ultrasound video frames as teacher input data and second training ultrasound data. PNG media_image2.png 959 1427 media_image2.png Greyscale Further, RUBIN teaches “ acquiring, for each teacher model of the plurality of the trained teacher models, corresponding first estimated data output from the teacher model, in response to input of first ultrasound data to the teacher model ;” See RUBIN in page 33, section 5.3. 3D-to-2D Attention Distillation for Videos, describes "Here, we utilized a 3D-DETR R50-6/6 model as the teacher network. The network applies a series of k=6 spatio-temporal 3D convolutions to a stack of ultrasound video frames. The effect is that the temporal dimension is convolved out to arrive at a 2D feature map with a singular temporal dimension, k=1, while image width and height remains unchanged. The learned feature map is then fed to the remainder of the network that has the same architecture as the 2D DETR-R50-6/6 teacher network." Here, RUBIN describes generating teacher models for both 2D and 3D ultrasound images. Additionally, RUBIN shows that for each teacher model, a corresponding feature map (i.e. image output ) was generated for that model. See Tables 1 and 2, and figure 2. PNG media_image2.png 959 1427 media_image2.png Greyscale PNG media_image3.png 908 1484 media_image3.png Greyscale RUBIN in figure 2 shows the ultrasound image output in form of results from decoder distillation, where the decoder distillation is involved in image data output, particularly in tasks involving image generation, restoration, and complex segmentation where a decoder network is used to reconstruct or generate the final pixel-level output. From tables 1 and 2, RUBIN describes there are two teacher models, one for 2D and a second for 3D, and they both have a corresponding image data output from their respective teacher model from input of an ultrasound image data (indicated from input frame(s) in figure 2). Further, RUBIN teaches “selecting a first particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models; ” See RUBIN in page 33, section 5.3. 3D-to-2D Attention Distillation for Videos, describes "Here, we utilized a 3D-DETR R50-6/6 model as the teacher network. The network applies a series of k=6 spatio-temporal 3D convolutions to a stack of ultrasound video frames. The effect is that the temporal dimension is convolved out to arrive at a 2D feature map with a singular temporal dimension, k=1, while image width and height remains unchanged. The learned feature map is then fed to the remainder of the network that has the same architecture as the 2D DETR-R50-6/6 teacher network." Here, RUBIN describes generating two teacher models for both 2D and 3D ultrasound images. Additionally, RUBIN shows that for each teacher model, a corresponding feature map (i.e. corresponding estimated image output ) was generated for that model. Further, RUBIN teaches “ and training a student model with the first ultrasound data as student input data and the corresponding first estimated data of the selected first particular teacher model as student target data .” See RUBIN in page 32, table 1, and in first paragraph in section 5.2. 2D-to-2D Attention Distillation for Images describe "Table1 summarizes model sizes, inference speeds, and average precision results for a series of student networks, and compares these to the DETR-R50 6/6 teacher model, a state-of-the-art Faster R-CNN model, and a baseline model trained without attention distillation. Here, we trained student models starting with single (1/1) encoder-decoder pairs up to 3/3. All student models were trained with the attention distillation hyper-parameter fixed to α=0.7. We can see that the student models approach the mAP50 and mAP50 short performance of the full teacher model while improving upon the processing frame rate. In particular, the DETR-R50-2/2 student model achieves a mAP50 of 0.643 vs.0.655 for the full teacher model, while increasing the frame rate from 26 to 43 FPS." Here, RUBIN shows that the teacher model has a corresponding student model DETR-R50-2/2, where the student model performs similar to the teacher model while also improving the processing of the image frame rate. The encoder-decoder pairs are closely related to corresponding image data, designed to map input images and return with an output image domain. Note since attention distillation is a method where a smaller student model is trained to mimic the patterns or information that a larger, pre-trained teacher model has, this helps improve the student model performance by teaching the model where to focus within the data, enhancing image detection tasks. Here, RUBIN teaches that the student model is trained with the parameters and alpha values, for their respective set of ultrasound data according to an initial teacher model (which was DETR-R50 6/6 in this case). See RUBIN in page 33, section 5.3 for more details. However, RUBIN did not teach: “A method, comprising: inputting first training ultrasound data, including a fundamental component and a harmonic component, … including the harmonic component, as teacher target data , ” “ selecting a first particular teacher model , of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models;” In an analogous art, LI teaches “A method, comprising: inputting first training ultrasound data, including a fundamental component and a harmonic component, … including the harmonic component, as teacher target data” See LI in paragraph [n0098] describe “ By modeling intravascular ultrasound images in polar coordinates, and using a dense distance regression network model, Euclidean distance and the location of the luminal centroid are predicted. During the training process of the network, knowledge distillation is used, with a dense distance regression network as the teacher model and a lightweight compressed network as the student model, to achieve knowledge transfer between the two.” Here, LI describes using a teacher model and student model for training, where the intravascular ultrasound images and predicted location of the luminal centroid is viewed as a target data for a teacher model. Further, see LI in [n0065] describe “This extracts the periodically changing high-frequency motion components in the motion offset to correct motion artifacts in intravascular ultrasound image sequences, making the motion correction effect more robust and improving the correction effect of motion artifacts .” LI applies the images to ultrasound data. Later, LI in [n0052] describe "Specifically, the input image is segmented using a student model after knowledge distillation, and data augmentation is performed using random flipping, random rotation, and discrete wavelet decomposition. The original image is enlarged to 512*512 pixels and then processed using discrete wavelet decomposition to become an input image of size 256*256 pixels with 4 channels. Each channel corresponds to the low-frequency information, vertical high-frequency information, horizontal high frequency information, and diagonal high-frequency information of the discrete wavelet processing result." Here in [n0052], LI describes using input images for student models after knowledge distillation, where this process (in [n0098]) refers to inputting ultrasound image data initially into a teacher model to later help train the student model to analyze the image data (i.e. input first training ultrasound data as teacher target data). Also in [n0052], LI discusses inputting an ultrasound image including low and high frequency discrete wavelet (i.e. which relate to fundamental and harmonic components) . Since discrete means whole number or integer multiple and since the harmonic frequencies are integer multiples of the fundamental frequency, LI relates to the fundamental and harmonic frequencies in ultrasound data. See LI in [n0087] for more details. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the base reference of RUBIN along with the secondary reference of LI by using the teachings of RUBIN for running a plurality of teacher models with corresponding estimated data outputs, with the teachings of LI of fundamental and harmonic components in ultrasound data. One of ordinary skill in the art would be motivated to do so because by integrating LI’s framework into the methods of RUBIN, one with ordinary skill in the art would achieve the goal of providing an “reduces the computational cost and improves the overall running speed of the algorithm while ensuring the accuracy of the segmentation effect,” (LI, [n0098]). However, RUBIN in view of LI, did not teach “ selecting a first particular teacher model , of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models;” In an analogous method, ZHANG teaches “ selecting a first particular teacher model , of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models;” See ZHANG in [n0011] describe "Depending on the specific optimization task, the selection of knowledge type, teacher-student neural network structure, and knowledge distillation algorithm can be comprehensively considered, and the optimal choice can be made based on the above three factors to achieve the best distillation performance"; Here, ZHANG discuss selection of a teacher-student neural network structure, which means selecting a teacher model, that can train the student model to achieve the best distillation performance . Further, see Zhang in [n0021] mention "The model building module is used to build a pre-trained teacher network model and an untrained first student network model." Further, see ZHANG in [n0090] describe "The parameters corresponding to the best-performing model are selected and used to adjust the model parameters; the test set is used to make predictions after the optimal model is obtained through the training and validation sets. It is used to measure the performance and classification ability of the optimal model." ZHANG describes a selection step of choosing a teacher model that helps to train a student model to learn data analysis of image data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the base reference of RUBIN along with the secondary reference of LI with the teachings of ZHANG by using the teachings of RUBIN and LI in inputting ultrasound data and training teacher and student models, with ZHANG’s teaching of selecting a teacher model. One of ordinary skill in the art would be motivated to do so because by integrating ZHANG’s framework into the methods of RUBIN and LI, one with ordinary skill in the art would achieve the goal of providing an “ultrasonic standard section image recognition network model obtained through training has the advantages of being high in recognition efficiency, light in structure and convenient to deploy” (ZHANG, abstract). Claim 2: Regarding claim 2, RUBIN in view of LI, further in view of ZHANG teaches the limitations of claim 1. Further, ZHANG teaches “ The method of claim 1, wherein each of the plurality of trained teacher models has a different model structure and more parameters than the student model .” See ZHANG in [n0041] describe "The teacher network model and the first student network model have similar structures, both being convolutional neural networks with residual connections. The difference is that the first student network model has fewer parameters and a lighter structure compared to the teacher network model" Here, ZHANG mentions that the student network model is structurally different compared to a teacher network model in terms of having fewer parameters and a lighter structure compared to the teacher model. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN and LI with the teachings of ZHANG by using the teachings of RUBIN and LI for inputting ultrasound data and training teacher and student models, with ZHANG’s teaching of teacher model has different model structure and more parameters than the student model. One of ordinary skill in the art would be motivated to do so because by integrating ZHANG’s framework into the methods of RUBIN and LI, one with ordinary skill in the art would achieve the goal of providing an “ultrasonic standard section image recognition network model obtained through training has the advantages of being high in recognition efficiency, light in structure and convenient to deploy” (ZHANG, abstract). Claim 6: Regarding claim 6, RUBIN teaches “ input first training ultrasound data , … to each of a plurality of teacher models, and train each teacher model of the plurality of teacher models with the first training ultrasound as teacher input data and second training ultrasound data, including the harmonic component, as teacher target data;” See RUBIN in page 33, section 5.3. 3D-to-2D Attention Distillation for Videos, describes "Here, we utilized a 3D-DETR R50-6/6 model as the teacher network. The network applies a series of k=6 spatio-temporal 3D convolutions to a stack of ultrasound video frames. The effect is that the temporal dimension is convolved out to arrive at a 2D feature map with a singular temporal dimension, k=1, while image width and height remains unchanged. The learned feature map is then fed to the remainder of the network that has the same architecture as the 2D DETR-R50-6/6 teacher network." Here, RUBIN describes generating teacher models for both 2D and 3D ultrasound images, indicating a plurality of teacher models. Additionally, RUBIN shows that for each teacher model, a corresponding feature map (i.e. image output ) was generated for that model. See Tables 1 and 2, and figure 2. RUBIN describes using the ultrasound video frames as teacher input data and second training ultrasound data. PNG media_image4.png 930 1383 media_image4.png Greyscale Further, RUBIN teaches “ acquire, for each teacher model of the plurality of the trained teacher models, corresponding first estimated data output from the teacher model, in response to input of first ultrasound data to the teacher model ;” See RUBIN in page 33, section 5.3. 3D-to-2D Attention Distillation for Videos, describes "Here, we utilized a 3D-DETR R50-6/6 model as the teacher network. The network applies a series of k=6 spatio-temporal 3D convolutions to a stack of ultrasound video frames. The effect is that the temporal dimension is convolved out to arrive at a 2D feature map with a singular temporal dimension, k=1, while image width and height remains unchanged. The learned feature map is then fed to the remainder of the network that has the same architecture as the 2D DETR-R50-6/6 teacher network." Here, RUBIN describes generating teacher models for both 2D and 3D ultrasound images. Additionally, RUBIN shows that for each teacher model, a corresponding feature map (i.e. image output ) was generated for that model. See Tables 1 and 2, and figure 2. PNG media_image3.png 908 1484 media_image3.png Greyscale RUBIN in figure 2 shows the ultrasound image output in form of results from decoder distillation, where the decoder distillation is involved in image data output, particularly in tasks involving image generation, restoration, and complex segmentation where a decoder network is used to reconstruct or generate the final pixel-level output. From tables 1 and 2, RUBIN describes there are two teacher models, one for 2D and a second for 3D, and they both have a corresponding image data output from their respective teacher model from input of an ultrasound image data (indicated from input frame(s) in figure 2). Further, RUBIN teaches “select a first particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models; ” See RUBIN in page 33, section 5.3. 3D-to-2D Attention Distillation for Videos, describes "Here, we utilized a 3D-DETR R50-6/6 model as the teacher network. The network applies a series of k=6 spatio-temporal 3D convolutions to a stack of ultrasound video frames. The effect is that the temporal dimension is convolved out to arrive at a 2D feature map with a singular temporal dimension, k=1, while image width and height remains unchanged. The learned feature map is then fed to the remainder of the network that has the same architecture as the 2D DETR-R50-6/6 teacher network." Here, RUBIN describes generating two teacher models for both 2D and 3D ultrasound images. Additionally, RUBIN shows that for each teacher model, a corresponding feature map (i.e. corresponding estimated image output ) was generated for that model. Further, RUBIN teaches “ and train a student model with the first ultrasound data as student input data and the corresponding first estimated data of the selected first particular teacher model as student target data .” See RUBIN in page 32, table 1, and in first paragraph in section 5.2. 2D-to-2D Attention Distillation for Images describe "Table1 summarizes model sizes, inference speeds, and average precision results for a series of student networks, and compares these to the DETR-R50 6/6 teacher model, a state-of-the-art Faster R-CNN model, and a baseline model trained without attention distillation. Here, we trained student models starting with single (1/1) encoder-decoder pairs up to 3/3. All student models were trained with the attention distillation hyper-parameter fixed to α=0.7. We can see that the student models approach the mAP50 and mAP50 short performance of the full teacher model while improving upon the processing frame rate. In particular, the DETR-R50-2/2 student model achieves a mAP50 of 0.643 vs.0.655 for the full teacher model, while increasing the frame rate from 26 to 43 FPS." Here, RUBIN shows that the teacher model has a corresponding student model DETR-R50-2/2, where the student model performs similar to the teacher model while also improving the processing of the image frame rate. The encoder-decoder pairs are closely related to corresponding image data, designed to map input images and return with an output image domain. Note since attention distillation is a method where a smaller student model is trained to mimic the patterns or information that a larger, pre-trained teacher model has, this helps improve the student model performance by teaching the model where to focus within the data, enhancing image detection tasks. Here, RUBIN teaches that the student model is trained with the parameters and alpha values, for their respective set of ultrasound data according to an initial teacher model (which was DETR-R50 6/6 in this case). See RUBIN in page 33, section 5.3 for more details. However, RUBIN did not teach: “ An apparatus, comprising: processing circuitry configured to …” “ input first training ultrasound data , including a fundamental component and a harmonic component, … including the harmonic component, as teacher target data ,” “ select a first particular teacher model , of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models;” In an analogous method, LI teaches “ An apparatus, comprising: processing circuitry configured , …” See LI in [n0014] describe "According to a third aspect, embodiments of the present invention also provide an electronic device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in the first aspect or any of its optional embodiments". Here, LI is showing a processor running computer instructions to implement any method, which teaches processing circuitry. Further, LI teaches “ to input first training ultrasound data , including a fundamental component and a harmonic component, … including the harmonic component, as teacher target data , ” See LI in paragraph [n0098] describe “ By modeling intravascular ultrasound images in polar coordinates, and using a dense distance regression network model, Euclidean distance and the location of the luminal centroid are predicted. During the training process of the network, knowledge distillation is used, with a dense distance regression network as the teacher model and a lightweight compressed network as the student model, to achieve knowledge transfer between the two.” Here, LI describes using a teacher model and student model for training, where the intravascular ultrasound images and predicted location of the luminal centroid is viewed as a target data for a teacher model. Further, see LI in [n0065] describe “This extracts the periodically changing high-frequency motion components in the motion offset to correct motion artifacts in intravascular ultrasound image sequences, making the motion correction effect more robust and improving the correction effect of motion artifacts .” LI applies the images to ultrasound data. Later, LI in [n0052] describe "Specifically, the input image is segmented using a student model after knowledge distillation, and data augmentation is performed using random flipping, random rotation, and discrete wavelet decomposition. The original image is enlarged to 512*512 pixels and then processed using discrete wavelet decomposition to become an input image of size 256*256 pixels with 4 channels. Each channel corresponds to the low-frequency information, vertical high-frequency information, horizontal high frequency information, and diagonal high-frequency information of the discrete wavelet processing result." Here in [n0052], LI describes using input images for student models after knowledge distillation, where this process (in [n0098]) refers to inputting initial ultrasound image data into a teacher model to later help train the student model to analyze the image data (i.e. input first training ultrasound data as teacher target data). First is construed by examiner to mean initial data. Also in [n0052], LI discusses inputting an ultrasound image including low and high frequency discrete wavelet (i.e. which relate to fundamental and harmonic components) . Since discrete means whole number or integer multiple and since the harmonic frequencies are integer multiples of the fundamental frequency, LI relates to the fundamental and harmonic frequencies in ultrasound data. See LI in [n0087] for more details. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the base reference of RUBIN along with the secondary reference of LI by using the teachings of RUBIN for running a plurality of teacher models with corresponding estimated data outputs, with the teachings of LI of fundamental and harmonic components in ultrasound data. One of ordinary skill in the art would be motivated to do so because by integrating LI’s framework into the methods of RUBIN, one with ordinary skill in the art would achieve the goal of providing an “reduces the computational cost and improves the overall running speed of the algorithm while ensuring the accuracy of the segmentation effect,” (LI, [n0098]). However, RUBIN in view of LI, did not teach “ select a first particular teacher model , of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models;” In an analogous method, ZHANG teaches “ select a first particular teacher model , of the plurality of trained teacher models, by evaluating the corresponding first estimated data output from each of the trained teacher models;” See ZHANG in [n0011] describe "Depending on the specific optimization task, the selection of knowledge type, teacher-student neural network structure, and knowledge distillation algorithm can be comprehensively considered, and the optimal choice can be made based on the above three factors to achieve the best distillation performance"; Here, ZHANG discuss selection of a teacher-student neural network structure, which means selecting a teacher model, that can train the student model to achieve the best distillation performance . Further, see Zhang in [n0021] mention "The model building module is used to build a pre-trained teacher network model and an untrained first student network model." Further, see ZHANG in [n0090] describe "The parameters corresponding to the best-performing model are selected and used to adjust the model parameters; the test set is used to make predictions after the optimal model is obtained through the training and validation sets. It is used to measure the performance and classification ability of the optimal model." ZHANG describes a selection step of choosing a teacher model that helps to train a student model to learn data analysis of image data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN and LI with the teachings of ZHANG by using the teachings of RUBIN and LI for inputting ultrasound data and training teacher and student models, with ZHANG’s teaching of selecting a teacher model. One of ordinary skill in the art would be motivated to do so because by integrating ZHANG’s framework into the methods of RUBIN and LI, one with ordinary skill in the art would achieve the goal of providing an “ultrasonic standard section image recognition network model obtained through training has the advantages of being high in recognition efficiency, light in structure and convenient to deploy” (ZHANG, abstract). Claim 7: Regarding claim 7, RUBIN in view of LI, further in view of ZHANG teaches the limitations of claim 6. Regarding claim 7, the claim recites similar limitations as corresponding claim 2 and is rejected for similar reasons as claim 2 using similar teachings and rationale . 07-21-aia AIA Claim s 3 and 8 are rejected under 35 U.S.C. 103 as being unpatentable over RUBIN, in view of LI, further in view of ZHANG, and further in view of Wang, Y. et al., (Pub. No. JP2023043825A), published on March 29, 2023, (hereafter, WANG), and further in view of Rubin, J. et al., (US PG Pub. US20240371500A1), filed July 20, 2022, (hereafter, RUBIN_22_patent) . Claim 3: Regarding claim 3, RUBIN in view of LI, further in view of ZHANG, teaches the limitations of claim 1. Further, ZHANG teaches “ and training the student model with the second ultrasound data as the student input data and the corresponding second estimated data of the selected second particular teacher model as the student target data .” See ZHANG in [n0019] describe "The second student network model is trained using the self-distillation knowledge of the second student network model and the similarity matrix prediction information between the output samples of the teacher network model and the second student network model to obtain a fetal ultrasound standard section image recognition network model". Here, ZHANG describes training a second student network model with output samples of the teacher network model (i.e. corresponding second estimated data of the selected second particular teacher model), and according to a translated figure for the student network model, the input images have an order x1, x2, x3, etc., and since there a variety of ultrasound images, this relates to second ultrasound data as the student input data. Note the examiner construes second to mean an initial or subsequent data. PNG media_image5.png 421 661 media_image5.png Greyscale Further, see ZHANG in [n0068] describe at step “S301, Input the dataset into the second student network model and the teacher network model to obtain the first output sample data and the first soft label data output by the second student network model, and the second output sample data and the second soft label data output by the teacher network model”. Here, ZHANG shows that a student model receives an input ultrasound imaging data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN and LI with the teachings of ZHANG by using the teachings of RUBIN and LI in inputting ultrasound data and training teacher and student models, with ZHANG’s teaching of training the student model with the second ultrasound data as the student input data and the corresponding second estimated data of the selected second particular teacher model as the student target data. One of ordinary skill in the art would be motivated to do so because by integrating ZHANG’s framework into the methods of RUBIN and LI, one with ordinary skill in the art would achieve the goal of providing an “ultrasonic standard section image recognition network model obtained through training has the advantages of being high in recognition efficiency, light in structure and convenient to deploy” (ZHANG, abstract). However, RUBIN in view of LI, further in view of ZHANG, did not teach “ The method of claim 1, further comprising: acquiring, for each teacher model of the plurality of the trained teacher models, corresponding second estimated data output from the teacher model, in response to input of second ultrasound data, different from the first ultrasound data, to the teacher model ;” “ selecting a second particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding second estimated data output from each of the trained teacher models ;” In an analogous method, WANG teaches “ The method of claim 1, further comprising: acquiring, for each teacher model of the plurality of the trained teacher models , corresponding second estimated data output from the teacher model, in response to input of second ultrasound data, different from the first ultrasound data, to the teacher model ;” See WANG in [0055] describes "At block 516, a processing unit implementing method 500 uses the first set of images and a portion of the object data determined from one or more outputs of the first machine learning model to generate A second machine learning model may be trained to detect objects of the target class in the second set. " Here, WANG shows that a portion of the data contains outputs from the first machine learning model, but not all of the data was the same, and shows the input of second imaging data in ultrasound is different than the first. Further, see WANG in [0060] describe "The target output may also include mask data associated with each object shown in each training image. Mask data may be obtained based on one or more outputs of the second machine learning model. In some examples, the second machine learning model may correspond to the teacher model described herein. For example, a set of training images may be provided as input to a second machine learning model. A second machine learning model detects, for a given input image, one or more objects of at least one of a plurality of classes represented in the given input image; and predicting at least mask data associated with each detected object. " Here, WANG shows that for a teacher model, a corresponding set of imaging data was acquired (i.e. second estimated data output ). See WANG in [0147] mention “the installation pipeline 1410 can include any number of applications, such as imaging devices, sequencing devices, genomics devices, including AI-assisted annotation, as described above. It may be applied continuously, non-continuously, or otherwise to imaging data (and/or other data types) generated by the device or the like. In at least one embodiment, the installation pipeline 1410 for individual devices includes a virtual machine for the device (e.g., virtual ultrasound machine , virtual CT scan machine, virtual sequence machine, virtual sequence machine, virtual ultrasound machine , virtual CT scan machine, virtual device, etc.)” . Here, WANG includes data to be ultrasound images. See [0169] for more information. Further, WANG teaches “selecting a second particular teacher model, of the plurality of trained teacher models , by evaluating the corresponding second estimated data output from each of the trained teacher models ;” See WANG in [0057] describe “the trained second machine learning model may be a multi-head machine learning model, as described above. In some embodiments, the processing unit may identify one or more heads of the second machine learning model corresponding to predicting mask data for a given input image; A second machine learning model may be updated to remove the plurality of identified heads. In some examples, the processing unit may provide the second set of images as input to the second machine learning model and obtain one or more outputs of the second machine learning model. The processing unit may determine additional object data associated with each of the second set of images from the one or more acquired outputs.” Here, WANG mentions corresponding to predicting mask data for a given input image relate to corresponding second estimated data output of a trained teacher model. See WANG in [0060] for details. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN, LI, and ZHANG, with the teachings of WANG by using the teachings of RUBIN, LI, and ZHANG, and incorporate with WANG’s teaching of evaluating the corresponding second estimated data output of a teacher model for ultrasound data. One of ordinary skill in the art would be motivated to do so because by integrating WANG’s framework into the methods of RUBIN, LI, and ZHANG, one with ordinary skill in the art would achieve the goal of a method that “becomes even more user-friendly, provides easier integration, and is more flexible. Able to produce accurate, efficient and timely results,” (WANG, [0148]). However, RUBIN in view of LI, further in view of ZHANG, further in view of WANG did not teach: “ The method of claim 1, further comprising: acquiring, for each teacher model of the plurality of the trained teacher models , …” “ selecting a second particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding second estimated data output from each of the trained teacher models;” In an analogous method, RUBIN_22_patent teaches “ The method of claim 1, further comprising: acquiring, for each teacher model of the plurality of the trained teacher models , …” See RUBIN_22_patent in [0060] describe “The method 100 may therefore leverage knowledge distillation from the second ML model in order to improve training and/or use of the first ML model. Knowledge distillation may be regarded as a type of model compression technique where a small student network attempts to extract encoded knowledge from single or multiple large teacher model(s).” Here, RUBIN_22_patent describes using multiple teacher models for the method. Further, RUBIN_22_patent in paragraph [0090] describe “The selection of the first ML model may occur if a small model needs to deployed, less accuracy is needed, a faster speed is needed and/or if the output of the first ML model is to be used in real-time. The selection of the second ML model may occur if a larger model needs to be deployed for higher accuracy and speed/no real-time usage is acceptable.” Here, RUBIN_22_patent describes selecting a first and a second ML model, where the ML model can be either a teacher model or a student model, from multiple teacher models or multiple student models. Further, RUBIN_22_patent teaches “ selecting a second particular teacher model, of the plurality of trained teacher models, by evaluating the corresponding second estimated data output from each of the trained teacher models;” See RUBIN_22_patent in [0060] describe “The method 100 may therefore leverage knowledge distillation from the second ML model in order to improve training and/or use of the first ML model. Knowledge distillation may be regarded as a type of model compression technique where a small student network attempts to extract encoded knowledge from single or multiple large teacher model(s).” Here, RUBIN_22_patent describes using multiple teacher models for the method. Further, RUBIN_22_patent in paragraph [0090] describe “The selection of the first ML model may occur if a small model needs to deployed, less accuracy is needed, a faster speed is needed and/or if the output of the first ML model is to be used in real-time. The selection of the second ML model may occur if a larger model needs to be deployed for higher accuracy and speed/no real-time usage is acceptable.” Here, RUBIN_22_patent describes selecting a first and a second ML model, where the ML model can be either a teacher model or a student model, from multiple teacher models or multiple student models. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN, LI, ZHANG, and WANG with the teachings of RUBIN_22_patent by using the teachings of RUBIN, LI, ZHANG, and WANG and incorporate with RUBIN_22_patent’s teaching of outputting ultrasound data by inputting input ultrasound data into the trained student model. One of ordinary skill in the art would be motivated to do so because by integrating RUBIN_22_patent’s framework into the methods of RUBIN, LI, ZHANG, and WANG, one with ordinary skill in the art would achieve the goal of a method that “provide[s] a machine learning model compression and/or distribution method to create fast and efficient image and video object detectors. Certain embodiments may allow small, fast and/or lightweight detector models to achieve up to the performance of more complex models (which may be large, slow and/or expensive) while still being able to run in real-time on resource-limited hardware,” (RUBIN_22_patent, [0062]). Claim 8: Regarding claim 8, RUBIN in view of LI, further in view of ZHANG, teaches the limitations of claim 6. Regarding claim 8, the claim recites similar limitations as corresponding claim 3 and is rejected for similar reasons as claim 3 using similar teachings and rationale . 07-21-aia AIA Claim s 4 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over RUBIN, in view of LI, further in view of ZHANG, further in view of WANG, and further in view of RUBIN_22_patent, and further in view of Hall, J. et al., (US PG Pub. No. US20220343178A1), published on October 27, 2022, (hereafter, HALL) . Claim 4: Regarding claim 4, RUBIN in view of LI, further in view of ZHANG, further in view of WANG, and further in view of RUBIN_22_patent, teaches the limitations of claim 3. However, RUBIN in view of LI, further in view of ZHANG, further in view of WANG, and further in view of RUBIN_22_patent, did not teach “ The method of claim 3, wherein the selected second particular teacher model is different from the selected first particular teacher model. ” In an analogous field, HALL teaches “ The method of claim 3, wherein the selected second particular teacher model is different from the selected first particular teacher model. ” See HALL in [0034] describe "In one form, the method may further comprise: generating a distillation model for each chromosomal group comprising: training a plurality of teacher models, wherein each of the plurality of teacher models is based on the best chromosomal group AI model for the respective group, and each of the plurality of teacher models is trained on at least part of the training dataset with a different set of initial conditions and image ordering ;" Here, HALL describes the process of training teacher models where each model contains a different set of initial conditions and image order, and shows that a second teacher model selected can be different from the initial teacher model selected for evaluating model performance. See HALL in [0011] for details on selecting a best model. Further, see HALL discuss in [0115] "To assist in further refinement of the AI model, pregnancy outcome data, such as detection (or not) of a heartbeat in the first ultrasound scan after implantation (normally around 6-10 weeks post fertilization), or aneuploidy results from a PGT-A test may be provided to the system. This allows the AI model to be retrained and updated as more data becomes available." This shows HALL applies this to ultrasound image data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN, LI, ZHANG, WANG, and RUBIN_22_patent and incorporate with the teachings of HALL by using the teachings of RUBIN, LI, ZHANG, WANG, and RUBIN_22_patent, with HALL’s teaching of a second teacher model is different from the first selected teacher model. One of ordinary skill in the art would be motivated to do so because by integrating HALL’s framework into the methods of RUBIN, LI, ZHANG, WANG, and RUBIN_22_patent, one with ordinary skill in the art would achieve the goal of “ensuring even distribution of images across test and training sets can be used to improve performance” (HALL, [0151]), and “reduce the complexity of the network and also reduce overfitting.” (HALL, [0170]). Claim 9: Regarding claim 9, RUBIN in view of LI, further in view of ZHANG, further in view of WANG, teaches the limitations of claim 8. Regarding claim 9, the claim recites similar limitations as corresponding claim 4 and is rejected for similar reasons as claim 4 using similar teachings and rationale . 07-21-aia AIA Claim s 5 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over RUBIN, in view of LI, further in view of ZHANG, further in view of Cai, Q. et al., in “The Application of Knowledge Distillation toward Fine-Grained Segmentation for Three-Vessel View of Fetal Heart Ultrasound Images,” published on July 14, 2022, available at https://onlinelibrary.wiley.com/doi/epdf/10.1155/2022/1765550, (hereafter, CAI) . Claim 5: Regarding claim 5, RUBIN in view of LI, further in view of ZHANG, teaches the limitations of claim 1. Further, LI teaches “ The method of claim 1, further comprising: generating output ultrasound data, including the harmonic component, by inputting input ultrasound data into the trained student model.” See LI in paragraph [n0052] describe "Specifically, the input image is segmented using a student model after knowledge distillation, and data augmentation is performed using random flipping, random rotation, and discrete wavelet decomposition. The original image is enlarged to 512*512 pixels and then processed using discrete wavelet decomposition to become an input image of size 256*256 pixels with 4 channels. Each channel corresponds to the low-frequency information, vertical high-frequency information, horizontal high frequency information, and diagonal high-frequency information of the discrete wavelet processing result." Here in [n0052], LI discusses inputting an ultrasound image including low and high frequency discrete wavelet (i.e. which relate to fundamental and harmonic components) . Since discrete means whole number or integer multiple and since the harmonic frequencies are integer multiples of the fundamental frequency, LI relates to the fundamental and harmonic frequencies in ultrasound data. However, RUBIN in view of LI, further in view of ZHANG, did not teach “The method of claim 1, further comprising: generating output ultrasound data, including the harmonic component , by inputting input ultrasound data into the trained student model .” In an analogous field, CAI teaches “The method of claim 1, further comprising: generating output ultrasound data , including the harmonic component , by inputting input ultrasound data into the trained student model .” See CAI in page 2, section 2.2 describes “After training teacher network with cropped inputs, its logits output is used to distill knowledge on channels for student model training. This allows the student model to have the same ability to segment the ROI as the teacher model. For the training of the student model, we use full-size images as input and align the logits of the teacher model to the full-size ones as additional knowledge. The loss of our network is made up of three components: PNG media_image6.png 23 207 media_image6.png Greyscale Here, ℒ CE ( ⋅ , ⋅ ) and ℒ Dice ( ⋅ , ⋅ ) denote cross-entropy loss and dice loss, respectively, y gt is the ground truth label mask, and y pred is the predicted probability map showing the probability that each pixel being categorized in each class.” See figure 1 for details. Here, CAI describes the logit outputs, which represent raw numerical output scores for each possible image class, from the final layer of a neural network in image processing tasks. Using these output scores, CAI also describes that the student network model is trained with data that was used to train the teacher network to also create output predictions on the ultrasound dataset, indicated by figure 1’s predictions and labels. See objectives section in page 1 and figure 2 from CAI for more information. PNG media_image7.png 358 750 media_image7.png Greyscale PNG media_image8.png 768 846 media_image8.png Greyscale It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of RUBIN, LI, and ZHANG, with the teachings of CAI by using the teachings of RUBIN, LI, and ZHANG, and incorporate with CAI’s teaching of outputting ultrasound data by inputting input ultrasound data into the trained student model. One of ordinary skill in the art would be motivated to do so because by integrating CAI’s framework into the methods of RUBIN, LI, and ZHANG, one with ordinary skill in the art would achieve the goal of a method that is "clinically important to collect and identify the three vessels, and the automatic segmentation algorithm could help doctors to easily operate and accurately diagnose CHD," (See ZHANG in pages 5-6, section 4. Discussion), and "achieved better results than DeepLabv3+ on all evaluation metrics, with performance gains of 1.8% on mean IoU (66.8% to 68.6%), 2.2% on mean PA (79.2% to 81.4%), and 1.2% on mean Dice (80.1% to 81.3%)," (see ZHANG in page 6, section 4. Discussion). Claim 10: Regarding claim 10, RUBIN in view of LI, further in view of ZHANG, teaches the limitations of claim 6. Regarding claim 10, the claim recites similar limitations as corresponding claim 5 and is rejected for similar reasons as claim 5 using similar teachings and rationale . 07-21-aia AIA Claim s 11, 15, 16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over LI, in view of WANG, and further in view of ZHANG . Claim 11: Regarding claim 11, LI further teaches “ A method, comprising: obtaining first ultrasound data, including a fundamental component and a harmonic component, as input ultrasound data and second ultrasound data, including the harmonic component, as target output ultrasound data corresponding to the first ultrasound data; ” See LI in paragraph [n0098] describe “During the training process of the network, knowledge distillation is used, with a dense distance regression network as the teacher model and a lightweight compressed network as the student model, to achieve knowledge transfer between the two.” LI describes using a teacher model and student model for training. See LI in [n0065] describe “This extracts the periodically changing high-frequency motion components in the motion offset to correct motion artifacts in intravascular ultrasound image sequences, making the motion correction effect more robust and improving the correction effect of motion artifacts .” LI applies the images to ultrasound data. Later, LI in [n0052] describe "Specifically, the input image is segmented using a student model after knowledge distillation, and data augmentation is performed using random flipping, random rotation, and discrete wavelet decomposition. The original image is enlarged to 512*512 pixels and then processed using discrete wavelet decomposition to become an input image of size 256*256 pixels with 4 channels. Each channel corresponds to the low-frequency information, vertical high-frequency information, horizontal high frequency information, and diagonal high-frequency information of the discrete wavelet processing result." Here in [n0052], LI describes using input images for student models after knowledge distillation, where this process (in [n0098]) refers to inputting initial ultrasound image data into a teacher model to later help train the student model to analyze the image data (i.e. input first training ultrasound data). First is construed by examiner to mean initial data. Also in [n0052], LI discusses inputting an ultrasound image including low and high frequency discrete wavelet (i.e. which relate to fundamental and harmonic components) . Since discrete means whole number or integer multiple and since the harmonic frequencies are integer multiples of the fundamental frequency, LI relates to the fundamental and harmonic frequencies in ultrasound data. Further, see LI in [n0065] describe “by performing the above steps, the vascular image correction method provided in this embodiment of the invention calculates the motion offset between the lumen contours of adjacent frames by utilizing the positions of the corresponding lumen centroids between adjacent frames, and calculates the cardiac cycle by utilizing the peak points of dissimilarity between adjacent frames. This extracts the periodically changing high-frequency motion components in the motion offset to correct motion artifacts in intravascular ultrasound image sequences, making the motion correction effect more robust and improving the correction effect of motion artifacts.” Here, LI describes the process of producing an image output of a reduced image size of the ultrasound image (i.e. second ultrasound data), and the target output is viewed is the motion corrected ultrasound image. See LI in [n0053] and [n0087] for more details. However, LI did not teach “ inputting the first ultrasound data to a previously trained teacher model to generate teacher output ultrasound data;” “inputting the first ultrasound data to a student model to generate student output ultrasound data;” “calculating a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data;” “and updating parameters of the student model based on the calculated loss value.” In an analogous field, WANG teaches “ inputting the first ultrasound data to a previously trained teacher model to generate teacher output ultrasound data;” See WANG in [0038] describe "A training data generator module 226 generates training data to be used to train a student object detection model to detect one or more objects associated with a target class for a given input image. obtain. Training data, in some examples, may include a set of training inputs and a set of target outputs. The set of training inputs may include one or more training images 276 provided as inputs to the trained teacher model 274 , as described above. The set of target outputs may include at least output image mask data 280 determined from one or more outputs of trained teacher model 274 ." Here, WANG describes inputting one or more training images (i.e. inputting first ultrasound data) to a teacher model 274, then the teacher model produces output image mask data 280 (i.e. generate teacher output ultrasound data). Further, see WANG in [0143] describe “In at least one embodiment, training system 1304 may execute a training pipeline 1404 similar to that described herein with respect to FIG. In at least one embodiment, if one or more machine learning models are to be used in deployment pipeline 1410 by deployment system 1306, training pipeline 1404 may include one or more (e.g., pre-trained ), and/or to implement one or more of the pretrained models 1406 (e.g., without the need for retraining or updating).” Here, WANG shows that the models used such as for teacher models, includes pre-trained models. See WANG in [0147] and [0169] further describe how this can be applied to ultrasound devices and imaging data. Further, WANG teaches “inputting the first ultrasound data to a student model to generate student output ultrasound data;” See WANG in [0055] describe "At block 516, a processing unit implementing method 500 uses the first set of images and a portion of the object data determined from one or more outputs of the first machine learning model to generate A second machine learning model may be trained to detect objects of the target class in the second set. A second machine learning model may be a student object detection model. The processing unit may train a student object detection model using training inputs and target outputs. The training input may include the first set of images (i.e., those provided as input to the trained teacher object detection model). The target output may include mask data associated with each detected object in the first set of images included in one or more outputs of the trained object detection model." Here, WANG shows that the student model takes inputs from a first set of images (i.e. inputting the first ultrasound data to a student model) and produces mask data associated with each detected object in the first set of images as target output (i.e. generate student output ultrasound data). See WANG in [0024] for more details. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the reference of LI with the teachings of WANG by using the teachings of LI and incorporate with WANG’s teaching of inputting respective teacher model or student model ultrasound data, and generating respective teacher or student output ultrasound data. One of ordinary skill in the art would be motivated to do so because by integrating WANG’s framework into the methods of LI would achieve a “more user-friendly, provides easier integration, and is more flexible. Able to produce accurate, efficient and timely results,” ([0148], WANG). However, LI in view of WANG did not teach “calculating a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data;” “and updating parameters of the student model based on the calculated loss value.” In an analogous field, ZHANG teaches “calculating a loss value of a loss function based on the generated teacher output ultrasound data, the generated student output ultrasound data, and the target output ultrasound data;” See ZHANG in [n0057] describe in step "S203, construct the first loss function based on the output of the hidden layer of the teacher network model and the output of the intermediate layer of the first student network model," Here, ZHANG teaches using output of information from both teacher and student models. Further, see ZHANG in [n0011] describes "Since the output of the last layer of neurons in a neural network is a probability distribution, the probability output distributions of the student network model and the teacher network model should be measured by cross-entropy, relative entropy, JS divergence , etc., and the loss function should be iterated continuously until it converges. Therefore, constructing a fetal standard section recognition method based on neural network knowledge distillation can extract prior knowledge information from high performance teacher neural network models, providing richer and more effective supervision information for network training. By incorporating an effective regularization mechanism, it can effectively realize automated neural network modeling for acquiring fetal standard ultrasound section images during prenatal ultrasound examinations." Here, ZHANG mentions the target output from the neural network processing on a fetal standard section recognition, and describes the loss values as cross-entropy, relative entropy, or JS divergence . See ZHANG in [n0050] and [n0068] for more details. Further, ZHANG teaches “and updating parameters of the student model based on the calculated loss value.” See ZHANG in [n0044] mention "Specifically, for training the teacher network model, the constructed deep residual neural network model is used to train the input data, and the process is iterated until the loss function converges and the global optimum is reached. The model with good performance is then saved as the teacher neural network model for subsequent steps." and [n0046] describe "This step provides prior knowledge based on the teacher network model for training the first student network model, which is equivalent to incorporating a new regularization mechanism to supervise the training of the student network model, resulting in better performance." Here, ZHANG mentions incorporating a new regularization mechanism to supervise training of the student network model, which relates regularization, a process that is designed to constrain or limit model complexity, to directly impact adjusting parameters during training (i.e. update parameters of student model ) . See ZHANG in [n0011] and [n0036] for more details. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of LI and WANG with the teachings of ZHANG by using the teachings of LI and WANG in inputting ultrasound data and training teacher and student models, with ZHANG’s teaching of calculating a loss value and updating parameters based on the calculated loss value. One of ordinary skill in the art would be motivated to do so because by integrating ZHANG’s framework into the methods of LI and WANG, one with ordinary skill in the art would achieve the goal of providing an “ultrasonic standard section image recognition network model obtained through training has the advantages of being high in recognition efficiency, light in structure and convenient to deploy, ” (ZHANG, abstract). Claim 15: Regarding claim 15, LI in view of WANG, and further in view of ZHANG teaches the limitations of claim 11. Further, LI teaches “The method of claim 11, further comprising: generating output ultrasound data, including the harmonic component , by inputting input ultrasound data into the trained student model.” See LI in paragraph [n0052] describe "Specifically, the input image is segmented using a student model after knowledge distillation, and data augmentation is performed using random flipping, random rotation, and discrete wavelet decomposition. The original image is enlarged to 512*512 pixels and then processed using discrete wavelet decomposition to become an input image of size 256*256 pixels with 4 channels. Each channel corresponds to the low-frequency information, vertical high-frequency information, horizontal high frequency information, and diagonal high-frequency information of the discrete wavelet processing result." Here in [n0052], Li discusses inputting an ultrasound image including low and high frequency wavelet (i.e. which relate to fundamental and harmonic component, since the harmonic frequencies are multiples of the fundamental frequency), and since the model is used for knowledge distillation. Further, WANG teaches “ The method of claim 11, further comprising: generating output ultrasound data , including the harmonic component, by inputting input ultrasound data into the trained student model. ” See WANG in [0048] describe "trained object detection model 322 (e.g., the trained student object detection model of FIG. 2) uses ROI data associated with a given input image, image It can be trained to predict mask data and/or image property data ." Here, WANG shows that given an input image (i.e. inputting input ultrasound data) into the trained student object detection model (i.e. trained student model) can be used to predict mask data and/or image property data. Further, see WANG [0011] describe "The second training data target output may further include ground truth data associated with each object detected in the set of images ." Here in [0011], WANG shows that the output of training data includes ground truth data related to the set of ultrasound images (i.e. generate output ultrasound data). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the reference of LI with the teachings of WANG by using the teachings of LI and incorporate with WANG’s teaching of inputting input ultrasound data into the trained student model. One of ordinary skill in the art would be motivated to do so because by integrating WANG’s framework into the methods of LI would achieve a “more user-friendly, provides easier integration, and is more flexible. Able to produce accurate, efficient and timely results,” ([0148], WANG). Claim 16: Regarding claim 16, LI further teaches “ An apparatus, comprising: processing circuitry configured …” See LI in [n0014] describe "According to a third aspect, embodiments of the present invention also provide an electronic device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in the first aspect or any of its optional embodiments". Here, LI is showing a processor running computer instructions to implement any method, which teaches processing circuitry. Regarding claim 16, the claim recites similar limitations as corresponding independent claim 11 and is rejected for similar reasons as claim 11 using similar teachings and rationale. Claim 20: Regarding claim 20, LI in view of WANG, and further in view of ZHANG teaches the limitations of claim 16. Regarding claim 20, the claim recites similar limitations as corresponding claim 15 and is rejected for similar reasons as claim 15 using similar teachings and rationale . 07-21-aia AIA Claim s 12 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over LI, in view of WANG, further in view of ZHANG, further in view of CAI, and further in view of Farooq, M. et al., in “Residual attention based uncertainty-guided mean teacher model for semi-supervised breast masses segmentation in 2D ultrasonography,” published on March 2023, available at https://www.sciencedirect.com/science/article/pii/S0895611122001434?via%3Dihub, (hereafter, FAROOQ) . Claim 12: Regarding claim 12, LI in view of WANG, further in view of ZHANG, teaches the limitations of claim 11. However, LI in view of WANG, further in view of ZHANG did not teach “ The method of claim 11, wherein the calculating step further comprises: calculating a first loss value of a first loss function based on the generated student output ultrasound data and the target output ultrasound data ;” “ calculating a second loss value of a second loss function based on the generated teacher output ultrasound data and generated student output ultrasound data ;” “ and calculating the loss value as a weighted sum of the first loss value and the second loss value .” In an analogous art, CAI teaches “ The method of claim 11, wherein the calculating step further comprises: calculating a first loss value of a first loss function based on the generated student output ultrasound data and the target output ultrasound data ;” See CAI in page 2, section 2.1 Channel-Wise Knowledge Distillation mention "Channel-wise knowledge distillation was first introduced by Zhou et al. [21]. Its basic idea is to convert the feature map on each channel to a probability map and then align the channel-wise probabilities of the teacher model and the student model. The corresponding channel-wise distillation loss ℒ C D( ⋅ , ⋅ ) is defined in the form of a KL divergence: PNG media_image9.png 192 979 media_image9.png Greyscale where T and S denote the teacher and student models". Here, CAI shows that the feature map is part of the output ultrasound data of both target output and student output. This feature map is incorporated into the distillation loss equation of KL divergence L KD , which uses the equation 1 as a function and calculates a loss quantity. Further, CAI teaches “ calculating a second loss value of a second loss function based on the generated teacher output ultrasound data and generated student output ultrasound data ;” See CAI in page 2, section 2.2 Our Method describe "As shown in Figure 1, the teacher and student networks use the same architecture. According to the ground truth label masks , we crop the region of interest (ROI, i.e., the region of three vessels) from full-size images to train the teacher model. Meanwhile, we record the coordinates of cropped regions to restore them to the corresponding spatial location in the full-size image afterwards. After training teacher network with cropped inputs, its logits output is used to distill knowledge on channels for student model training. This allows the student model to have the same ability to segment the ROI as the teacher model. For the training of the student model, we use full-size images as input and align the logits of the teacher model to the full-size ones as additional knowledge. The loss of our network is made up of three components:" PNG media_image6.png 23 207 media_image6.png Greyscale Here, CAI describes L ce and L dice shows second and third loss values, respectively, calculated from the equation 3 (i.e. loss function), that are part of the network of teacher and student models. The logit outputs along with ground-truth label masks are part of the teacher output ultrasound data and generated student output ultrasound data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of LI, WANG, and ZHANG, with the teachings of CAI by using the teachings of LI, WANG, and ZHANG, and incorporate with CAI’s teaching of first and second loss value and loss functions from respective output ultrasound data and generated student output ultrasound data. One of ordinary skill in the art would be motivated to do so because by integrating CAI’s framework into the methods of LI, WANG, and ZHANG, one with ordinary skill in the art would achieve the goal of “employ the channel-wise knowledge distillation method, as it was shown to be more effective than the pixel-wise knowledge distillation that simply aligns point-wise classification scores per pixel,” (CAI, page 2, section 2. Methods, part 2.1). However, LI in view of WANG, further in view of ZHANG, further in view of CAI, did not teach “ and calculating the loss value as a weighted sum of the first loss value and the second loss value .” In an analogous method, FAROOQ teaches “ and calculating the loss value as a weighted sum of the first loss value and the second loss value .” See FAROOQ in page 3, section I. Introduction, middle of first paragraph describe “The total loss has been calculated using the weighted sum of two loss functions that are supervised loss and uncertainty-guided consistency loss. Whereas, the evaluation of the consistency between the teacher and student model is calculated on the base of the uncertainty-guided consistency loss.” Here, FAROOQ explicitly teaches that a loss value is based on a weighted sum of at least two loss functions, and are part of a first loss value and a second loss value calculated from the teacher model and student model. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of LI, WANG, ZHANG, and CAI, with the teachings of FAROOQ by using the teachings of LI, WANG, ZHANG, and CAI, and incorporate with FAROOQ’s teaching of calculating a loss value as a weighted sum of the first loss value and the second loss value. One of ordinary skill in the art would be motivated to do so because by integrating FAROOQ’s framework into the methods of LI, WANG, ZHANG, and CAI, one with ordinary skill in the art would achieve “by adding residual attention module in the down-sampling and up-sampling part of the network to improve the feature extraction ability of the framework. …Significantly improved the segmentation accuracy by implying an uncertainty loss function between the predictions of the teacher and student networks,” (FAROOQ, page 3, first and second listed points). Claim 17: Regarding claim 17, LI in view of WANG, further in view of ZHANG, teaches the limitations of claim 16. Regarding claim 17, the claim recites similar limitations as corresponding claim 12 and is rejected for similar reasons as claim 12 using similar teachings and rationale . 07-21-aia AIA Claim s 13 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over LI, in view of WANG, further in view of ZHANG, and further in view of Anand, D. et al., (US PG Pub. No. US20230111306A1), published on April 13, 2023, (hereafter, ANAND) . Claim 13: Regarding claim 13, LI in view of WANG, further in view of ZHANG, teaches the limitations of claim 11. However, LI in view of WANG, further in view of ZHANG did not teach “ The method of claim 11, further comprising repeating the obtaining, inputting, inputting, calculating, and updating steps for different input ultrasound data and corresponding different second ultrasound data until the parameters of the student model satisfy a convergence criteria .” In an analogous art, ANAND teaches “ The method of claim 11, further comprising repeating the obtaining, inputting, inputting, calculating, and updating steps for different input ultrasound data and corresponding different second ultrasound data until the parameters of the student model satisfy a convergence criteria ,” See ANAND in [0061-0062] describe “The self-supervised aspect of this training process is based on comparing the feature representation outputs of the student and teacher networks, calculating loss based on agreement at 316, and iteratively updating the parameters of both the student and teacher networks at 318 until the feature representations generated by each of the student and teacher for the different image versions converge on the same feature representation (e.g., to a defined degree of similarity). The loss function used at 316 can vary. In some implementations, the loss can be based on distillation loss. In other implementations, the loss can be based on contrastive loss. The parameter updating at 318 can involve updating the parameters of both models by exotically moving the average of both the student and teacher model parameters. The final converged feature representation generated by the respective student and teacher transformer networks can be used as the representative feature representation for the input image 302.” Here, ANAND describes calculate loss + iteratively (repeatedly) update parameters from calculating loss based on agreement at step 316, and updating the different image versions (i.e. different second ultrasound data) until parameters of student and teacher networks converge to a defined degree of similarity (i.e. update parameters of student model until model satisfy a convergence criteria). Note the examiner construes updating parameters of a student model includes updating the parameters of both student and teacher models. Further, see ANAND in [0044] describe "The image-based inference output can include a data object that can be rendered (e.g., a visual data object ), stored , used as input for another processing task, or the like." ANAND here describes inputting the data. Further, See ANAND in [0070] for "For example, the selection criteria (defined by the paired view generation schema 112) may instruct the selection component to select one or more similar images to the training image that were acquired of the same anatomical region with the same capture modality yet with varying (e.g., withing a defined degree of similarity) acquisition parameters" Here, ANAND describes acquiring data, which relate to obtaining input data. See ANAND in [0045], [0069] for including ultrasound images as part of image data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of LI, WANG, and ZHANG, with the teachings of ANAND by using the teachings of LI, WANG, and ZHANG, and incorporate with ANAND’s teaching of outputting ultrasound data by inputting input ultrasound data into the trained student model. One of ordinary skill in the art would be motivated to do so because by integrating ANAND’s framework into the methods of LI, WANG, and ZHANG, one with ordinary skill in the art would achieve the goal of a method where “the training consists of iterative methods using numerical, optimization techniques that reduce the error between the desired class label and the algorithm's prediction. The newly trained model is then given new data as an input and, if trained well, can classify or otherwise provide assessment of novel data,” (See ANAND [0002]). Claim 18: Regarding claim 18, LI in view of WANG, further in view of ZHANG, teaches the limitations of claim 16. Regarding claim 18, the claim recites similar limitations as corresponding claim 13 and is rejected for similar reasons as claim 13 using similar teachings and rationale . 07-21-aia AIA Claim s 14 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over LI, in view of WANG, further in view of ZHANG, and further in view of Zou, W. et al., in "Graph Flow: Cross-Layer Graph Flow Distillation for Dual Efficient Medical Image Segmentation," published on November 24, 2022, available on https://ieeexplore.ieee.org/abstract/document/9963578 , (hereafter, ZOU) . Claim 14: Regarding claim 14, LI in view of WANG, further in view of ZHANG, teaches the limitations of claim 11. However, LI in view of WANG, further in view of ZHANG did not teach “ The method of claim 11, wherein the updating step comprises updating the parameters of the student model without updating parameters of the trained teacher model .” In an analogous art, ZOU teaches “ The method of claim 11, wherein the updating step comprises updating the parameters of the student model without updating parameters of the trained teacher model .” See ZOU in page 1164, section C. Adversarial Distillation note "Next, the student network S approximates the output distribution of the teacher network T as much as possible by maximizing L adv . " Further, see ZOU in page 1164 section F: Objective functions describe “In the optimization of our proposed Graph Flow, the training procedures can be implemented through three steps. At first, we employ L rec to train Pm while the parameters of pre-trained T are frozen. Next, the discriminator D is trained to discriminate Y s and Y t by minimizing L adv . At last, we optimize the student network S with an overall objective function L that contains the basic cross-entropy loss L ce , variation graph loss L vg , adversarial loss L g , and logits distillation loss L kd . The L can be presented as:” PNG media_image10.png 120 842 media_image10.png Greyscale Here, ZOU shows that the student network model S is optimized (and also updated) with an overall objective function L, while the parameters of pre-trained teacher network model T are frozen. Note the examiner construes updating to mean any form of changing information such as optimizing the model. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of LI, WANG, and ZHANG, with the teachings of ZOU by using the teachings of LI, WANG, and ZHANG, and incorporate with ZOU’s teaching of updating the parameters of the student model without updating parameters of the trained teacher model. One of ordinary skill in the art would be motivated to do so because by integrating ZOU’s framework into the methods of LI, WANG, and ZHANG, one with ordinary skill in the art would achieve the goal of achieving a “knowledge distillation method can greatly improve the performance of such networks without any extra parameters. Especially, ENet, the most lightweight network, obtains only 1.56% lower than FANet in ACC for gastric cancer pathology image segmentation. Moreover, on CVC-ClinicDB, our method can improve the performance of ENet by 3.92% in average DSC and 3.41% in mIOU, when teacher networks is Transformer architecture,” (see ZOU in page 1167, section 3. results of the Generalization Ablation). Claim 19: Regarding claim 19, LI in view of WANG, further in view of ZHANG, teaches the limitations of claim 16. Regarding claim 19, the claim recites similar limitations as corresponding claim 14 and is rejected for similar reasons as claim 14 using similar teachings and rationale. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to WENWEI ZENG whose telephone number is (571)272-7111. The examiner can normally be reached Monday-Friday, 8am-5pm. Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Usmaan Saeed can be reached at (571) 272-4046. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000. /WenWei Zeng/Examiner, Art Unit 2146 /USMAAN SAEED/Supervisory Patent Examiner, Art Unit 2146 Application/Control Number: 18/361,238 Page 2 Art Unit: 2146 Application/Control Number: 18/361,238 Page 3 Art Unit: 2146 Application/Control Number: 18/361,238 Page 4 Art Unit: 2146 Application/Control Number: 18/361,238 Page 5 Art Unit: 2146 Application/Control Number: 18/361,238 Page 6 Art Unit: 2146 Application/Control Number: 18/361,238 Page 7 Art Unit: 2146 Application/Control Number: 18/361,238 Page 8 Art Unit: 2146 Application/Control Number: 18/361,238 Page 9 Art Unit: 2146 Application/Control Number: 18/361,238 Page 10 Art Unit: 2146 Application/Control Number: 18/361,238 Page 11 Art Unit: 2146 Application/Control Number: 18/361,238 Page 12 Art Unit: 2146 Application/Control Number: 18/361,238 Page 13 Art Unit: 2146 Application/Control Number: 18/361,238 Page 14 Art Unit: 2146 Application/Control Number: 18/361,238 Page 15 Art Unit: 2146 Application/Control Number: 18/361,238 Page 16 Art Unit: 2146 Application/Control Number: 18/361,238 Page 17 Art Unit: 2146 Application/Control Number: 18/361,238 Page 18 Art Unit: 2146 Application/Control Number: 18/361,238 Page 19 Art Unit: 2146 Application/Control Number: 18/361,238 Page 20 Art Unit: 2146 Application/Control Number: 18/361,238 Page 21 Art Unit: 2146 Application/Control Number: 18/361,238 Page 22 Art Unit: 2146 Application/Control Number: 18/361,238 Page 23 Art Unit: 2146 Application/Control Number: 18/361,238 Page 24 Art Unit: 2146 Application/Control Number: 18/361,238 Page 25 Art Unit: 2146 Application/Control Number: 18/361,238 Page 26 Art Unit: 2146 Application/Control Number: 18/361,238 Page 27 Art Unit: 2146 Application/Control Number: 18/361,238 Page 28 Art Unit: 2146 Application/Control Number: 18/361,238 Page 29 Art Unit: 2146 Application/Control Number: 18/361,238 Page 30 Art Unit: 2146 Application/Control Number: 18/361,238 Page 31 Art Unit: 2146 Application/Control Number: 18/361,238 Page 32 Art Unit: 2146 Application/Control Number: 18/361,238 Page 33 Art Unit: 2146 Application/Control Number: 18/361,238 Page 34 Art Unit: 2146 Application/Control Number: 18/361,238 Page 35 Art Unit: 2146 Application/Control Number: 18/361,238 Page 36 Art Unit: 2146 Application/Control Number: 18/361,238 Page 38 Art Unit: 2146 Application/Control Number: 18/361,238 Page 39 Art Unit: 2146 Application/Control Number: 18/361,238 Page 40 Art Unit: 2146 Application/Control Number: 18/361,238 Page 41 Art Unit: 2146 Application/Control Number: 18/361,238 Page 42 Art Unit: 2146 Application/Control Number: 18/361,238 Page 43 Art Unit: 2146 Application/Control Number: 18/361,238 Page 44 Art Unit: 2146 Application/Control Number: 18/361,238 Page 45 Art Unit: 2146 Application/Control Number: 18/361,238 Page 46 Art Unit: 2146 Application/Control Number: 18/361,238 Page 47 Art Unit: 2146 Application/Control Number: 18/361,238 Page 48 Art Unit: 2146 Application/Control Number: 18/361,238 Page 49 Art Unit: 2146 Application/Control Number: 18/361,238 Page 50 Art Unit: 2146 Application/Control Number: 18/361,238 Page 51 Art Unit: 2146 Application/Control Number: 18/361,238 Page 52 Art Unit: 2146 Application/Control Number: 18/361,238 Page 53 Art Unit: 2146 Application/Control Number: 18/361,238 Page 54 Art Unit: 2146 Application/Control Number: 18/361,238 Page 55 Art Unit: 2146 Application/Control Number: 18/361,238 Page 56 Art Unit: 2146 Application/Control Number: 18/361,238 Page 57 Art Unit: 2146 Application/Control Number: 18/361,238 Page 58 Art Unit: 2146 Application/Control Number: 18/361,238 Page 59 Art Unit: 2146 Application/Control Number: 18/361,238 Page 60 Art Unit: 2146 Application/Control Number: 18/361,238 Page 61 Art Unit: 2146 Application/Control Number: 18/361,238 Page 62 Art Unit: 2146 Application/Control Number: 18/361,238 Page 63 Art Unit: 2146 Application/Control Number: 18/361,238 Page 64 Art Unit: 2146 Application/Control Number: 18/361,238 Page 65 Art Unit: 2146 Application/Control Number: 18/361,238 Page 66 Art Unit: 2146 Application/Control Number: 18/361,238 Page 67 Art Unit: 2146 Application/Control Number: 18/361,238 Page 68 Art Unit: 2146 Application/Control Number: 18/361,238 Page 69 Art Unit: 2146 Application/Control Number: 18/361,238 Page 70 Art Unit: 2146 Application/Control Number: 18/361,238 Page 71 Art Unit: 2146 Application/Control Number: 18/361,238 Page 72 Art Unit: 2146 Application/Control Number: 18/361,238 Page 73 Art Unit: 2146 Application/Control Number: 18/361,238 Page 74 Art Unit: 2146 Application/Control Number: 18/361,238 Page 75 Art Unit: 2146 Application/Control Number: 18/361,238 Page 76 Art Unit: 2146
