DETAILED ACTION
Notice of Pre-AIA or AIA Status
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
The information disclosure statements (IDS) submitted on 01/02/2025 was considered by the Examiner and entered into the record. However, it is impractical for the Examiner to review the references thoroughly given the number of references cited in this IDSs. A total of sixty-four (64) references spanning as many as fifty-five (55) pages were cited. By initialing each of the cited references on the accompanying 1449 forms, or by not striking through the cited reference, the Examiner is merely acknowledging the submission of the cited references and indicating that only a cursory review has been made of each cited reference. MPEP § 2004.13 states (emphasis added):
"It is desirable to avoid the submission of long lists of documents if it can be avoided. Eliminate clearly irrelevant and marginally pertinent cumulative information. If a long list is submitted, highlight those documents which have been specifically brought to applicant's attention and/or are known to be of most significance. See Penn Yan Boats, Inc. v. Sea Lark Boats, Inc., 359 F. Supp.948, 175 USPQ 260 (S.D. Fla. 1972), aft'd, 479 F.2d 1338, 178 USPQ 577 (5th Cir. 1973), cert. denied, 414 U.S. 874 (1974). But cf. Molins PLC v. Textron Inc., 48 F.3d 1172, 33 USPQ2d 1823(Fed. Cir. 1995)."
Further, it should be noted that an Applicant's duty of disclosure of material and information is not satisfied by presenting an Examiner with "a mountain of largely irrelevant [material] from which he is presumed to have been able, with his experience and with adequate time, to have found the critical [material]. It ignores the real world conditions under which Examiners work." Rohm & Haas Co. v. Crystal Chemical co., 722 F.2d 1556, 1573 [220 USPQ 289] (Fed. Cir.1983), cert. Denied, 469 U.S. 851 (1984). Patent Applicant has a duty not just to disclose pertinent prior art references but to make a disclosure in such a way as not to "bury" it within other disclosures of less relevant prior art; see Golden Valley Microwave Foods Inc. v. Weaver Popcorn Co. Inc., 24 USPQ2d 1801 (N.D. Ind. 1992); Molins PLC v. Textron Inc., 26 USPQ2d 1889, at 1899 (D.Del 1992); Penn Yan Boats, Inc. v. Sea Lark Boats, Inc. et al., 175 USPQ 260, at 272 (S.D. FI. 1972).
Given the large number of references cited on the IDS, the Examiner respectfully requests the cooperation of the Applicant in providing a concise explanation of relevance, such as the pertinent paragraphs, columns and line numbers, or drawings, which have caused each corresponding item to be listed on the IDS, since such action will ensure that information pertinent to the validity of any issued patent will not be overlooked.
Claim Rejections - 35 USC § 103
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 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.
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 of this title, 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.
Claims 1-5, 7, 9, 10-13, 15-17 and 19-24 are rejected under 35 U.S.C. 103 as being unpatentable over WANG et al. (US 20220229140 A1), hereinafter referenced as WANG in view of CORREIA et al. (US 20220373631 A1), hereinafter referenced as CORREIA.
Regarding claim 1, WANG explicitly teaches a method for generating magnetic resonance imaging (MRI) quantitative parameter maps (Fig. 34. Abstract-WANG discloses exemplary methods for quantitative mapping of physical properties, systems and computer-accessible medium can be provided to generate images of tissue magnetic susceptibility, transport parameters and oxygen consumption from magnetic resonance imaging data using the Bayesian inference approach. The structure prior knowledge can be characterized from known anatomic images using image feature extraction operation or artificial neural network. According to the embodiment, system, method and computer-accessible medium can be provided for determining physical properties associated with at least one structure), the method comprising:
receiving at least one multi-contrast magnetic resonance (MR) image of a subject (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. In paragraph [0233]-WANG discloses the process 800 begins by acquiring MR data corresponding to a subject (step 810). The MR data can correspond to a patient (e.g., the entirety of a patient or a portion of a patient, such as a particular portion to the patient's body). In paragraph [0248]-WANG discloses the MR data can be acquired using an MRI scanner using one or more suitable pulse sequences sensitizing multiple contrasts in tissue (step 822) (wherein MR data can be acquired using various parameters including a T1 weighted image dataset with a small fraction for T2 weighted, T2 FLAIR and diffusion weighted image datasets). Please also read paragraph [0136-0137]);
providing the at least one multi-contrast MR image of the subject to an artifact suppression deep learning network of a two-stage deep learning network (Fig. 33. Paragraph [0120]-WANG discloses the fidelity imposed network edit (FINE) is an artificial neural network, including convolutional neural network. FINE can achieve superior performance in quantitative susceptibility mapping (QSM) and undersampled multi-contrast reconstruction in MRI. Please also read paragraph [0136-0137, 0232, 0237, 0239, and 0249-0250]);
generating at least one multi-contrast MR image with suppressed undersampling artifacts using the artifact suppression deep learning network to suppress undersampling artifacts in the at least one multi-contrast MR image of the subject (Fig. 33. Paragraph [0136]-WANG discloses FINE was applied to multi-contrast MRI reconstruction with under-sampled data. In order to accelerate the time-consuming acquisition of certain contrasts, such as T2 weighted (T2w) or T2 Fluid Attenuated Inversion Recovery (T2FLAIR) images, k-space was under-sampled, thus requiring a regularized algorithm to recover images with minimal artifact. Please also read paragraph [0137-0139, 0146, 0237, 0239 and 0249-0250]);
providing the at least one multi-contrast MR image with suppressed undersampling artifacts to a parameter mapping deep learning network of the two-stage deep learning network (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to transform magnetic resonance (MR) signal data corresponding to a subject into multiple images that quantitatively depict the structure and/or composition and/or function of the subject. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain time-resolved MR data corresponding a subject as a contrast agent passing through tissue in an organ of the subject, and process this time-resolved MR data to generate a quantitative transport map of the subject (wherein a two-stage deep learning network is used for suppression of undersampling artifacts and quantitative parameter mapping). Please also read paragraph [0136-0137, 0237, 0239, 0249-0250]);
generating at least one quantitative MR parameter map based on the at least one multi-contrast MR image with suppressed undersampling artifacts using the parameter mapping deep learning network (Fig. 33. Paragraph In paragraph [0238]-WANG discloses the process 800 continues by determining prior knowledge about tissue magnetic susceptibility distribution (step 850) (wherein R2* is used to estimate susceptibility distribution). In paragraph [0239]-WANG discloses the process 800 continues by estimating a magnetic susceptibility distribution of the subject based, at least in part, on the prior information and data noise property (step 860). This can be determined using an artificial neural network. Please also read paragraph [0136, 0236, 0244-0247 and 0248-0251]); and
displaying at least one of the at least one multi-contrast MR image with suppressed undersampling artifacts, the at least one quantitative MR parameter map, and the corresponding uncertainty estimation map on a display (Fig. 33. Paragraph [0232]-WANG discloses using these quantitative maps of physical properties and multiple contrasts, one or more images of the subject can be generated and displayed to a user. The user can then use these images for diagnostic, therapeutic or experimental purposes, for example to investigate the structure and/or composition and/or function of the subject, and/or to diagnose various conditions or diseases based, and/or to treat various conditions or diseases based, at least in part, on the images. Please also read paragraph [0240, 0247 and 0251]).
WANG fails to explicitly teach generating an uncertainty estimation map for the at least one quantitative MR parameter map using the parameter mapping deep learning network; and
However, CORREIA explicitly teaches generating an uncertainty estimation map for the at least one quantitative MR parameter map using the parameter mapping deep learning network (Fig. 5. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data. In paragraph [0133]-CORREIA discloses deep learning can be used to reconstruct motion corrected tracer-kinetic maps. A potential solution to accelerate reconstruction time is to use a deep learning approach to directly estimate TK parameter maps from undersampled FPP-CMR data (DIREQT-NET). In paragraph [0134]-CORREIA discloses one approach to solve Eq. (5) using deep learning is to directly learn the nonlinear mapping between the undersampled k-space data d or aliased zero-filled undersampled reconstruction and the fully-sampled TK parameter maps using a deep neural network (e.g. convolutional neural network, CNN). In paragraph [0140]-CORREIA discloses other deep neural networks than CNNs can be used for DIREQT-NET, such as a Bayesian neural networks. In paragraph [0141]-CORREIA discloses the use of a Bayesian neural network may additionally provide uncertainty maps which may be useful in assessing the accuracy of the motion corrected tracer-kinetic map); and
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG having a method for generating magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having generating an uncertainty estimation map for the at least one quantitative MR parameter map using the parameter mapping deep learning network.
Wherein having WANG’s method having generating an uncertainty estimation map for the at least one quantitative MR parameter map using the parameter mapping deep learning network; and
The motivation behind the modification would have been to obtain a method that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 2, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG further teaches wherein one or more of the at least one multi-contrast MR image and at least one multi-contrast MR image with suppressed undersampling artifacts are multi-echo MR images (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. Please also read paragraph [0131 and 0136]).
Regarding claim 3, WANG in view of CORREIA explicitly teach he method according to claim 1, WANG further teaches wherein the artifact suppression learning network is a convolutional neural network (Fig. 33. Paragraph [0120]-WANG discloses described is an example of fidelity imposed network edit (FINE) where the network is an artificial neural network in deep learning, including a convolutional neural network. The experiments demonstrated that FINE can achieve superior performance in quantitative susceptibility mapping (QSM) and undersampled multi-contrast reconstruction in MRI. In paragraph [0146]-WANG discloses QSMs reconstructed by FINE are displayed in FIG. 18c. Severe shadow artifact were observed in DL and DLL2. These artifacts were markedly suppressed in FINE. Please also read paragraph [0232]).
Regarding claim 4, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG further teaches wherein the parameter mapping deep learning network is a convolutional neural network (Fig. 33. Paragraph [0120]-WANG discloses described is an example fidelity imposed network edit (FINE) where the network is an artificial neural network in deep learning, including a convolutional neural network. The experiments demonstrated that FINE can achieve superior performance in quantitative susceptibility mapping (QSM) and undersampled multi-contrast reconstruction in MRI. Further in paragraph [0232]-WANG discloses one or more of the above quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33).
Regarding claim 5, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG further teaches wherein the at least one multi-contrast MR image is a plurality of multi-contrast MR images reconstructed from undersampled k-space data (Fig. 33. Paragraph [0120]-WANG discloses the experiments demonstrated that FINE can achieve superior performance in undersampled multi-contrast reconstruction in MRI. In paragraph [0136]-WANG discloses FINE was applied to multi-contrast MRI reconstruction with under-sampled data. In order to accelerate the time-consuming acquisition of certain contrasts, such as T2 weighted (T2w) or T2 Fluid Attenuated Inversion Recovery (T2FLAIR) images, k-space was under-sampled, thus requiring a regularized algorithm to recover images with minimal artifact. In paragraph [0232]-WANG discloses the process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images).
Regarding claim 7, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG further teaches wherein the at least one quantitative MR parameter map is a plurality of quantitative MR parameter maps, wherein each quantitative MR parameter map corresponds to a different quantitative parameter (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject (wherein R2* is used to estimate susceptibility). The process 800 can be used to obtain time-resolved MR data corresponding a subject as a contrast agent passing through tissue in an organ of the subject, and process this time-resolved MR data to generate a quantitative transport map of the subject. Please also read paragraph [0245-0246]).
Regarding claim 9, WANG in view of CORREIA explicitly teach the method according to claim 7, WANG fails to explicitly teach wherein each quantitative MR parameter map in the plurality of quantitative MR parameter maps has a corresponding uncertainty estimation map.
However, CORREIA explicitly teaches wherein each quantitative MR parameter map in the plurality of quantitative MR parameter maps (Fig. 5. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data (wherein DIREQT can be combined with motion correction techniques to provide accurate quantitative maps from highly accelerated free-breathing data). In paragraph [0133]-CORREIA discloses Deep learning can be used to reconstruct motion corrected tracer-kinetic maps. A solution to accelerate reconstruction time is deep learning to directly estimate TK parameter maps from undersampled FPP-CMR data (DIREQT-NET). In paragraph [0135]-CORREIA discloses the trained CNN can be used to generate artefact-free TK parameter maps from undersampled FPP-CMR data. The network is trained to learn the mapping between the undersampled k-space data and TK parameter maps and output an estimate of residual maps. Please also read paragraph [0134]) has a corresponding uncertainty estimation map (Fig. 5. Paragraph [0140]-CORREIA discloses other deep neural networks than CNNs can be used for DIREQT-NET, such as a Bayesian neural networks. In paragraph [0141]-CORREIA discloses the use of a Bayesian neural network may additionally provide uncertainty maps which may be useful in assessing the accuracy of the motion corrected tracer-kinetic map).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA of having a method for generating magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having wherein each quantitative MR parameter map in the plurality of quantitative MR parameter maps has a corresponding uncertainty estimation map.
Wherein having WANG’s method having wherein each quantitative MR parameter map in the plurality of quantitative MR parameter maps has a corresponding uncertainty estimation map.
The motivation behind the modification would have been to obtain a method that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 10, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG fails to explicitly teach wherein the two-stage deep learning network is trained using a loss function that comprises a MR physics loss term.
However, CORREIA explicitly teaches wherein the two-stage deep learning network (Fig. 5. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data (wherein DIREQT-NET may be convolutional neural networks or Bayesian neural networks with uncertainty maps). In paragraph [0131]-CORREIA discloses in order to obtain accurate quantitative maps, the first-pass data must be motion compensated to minimise for the resulting artefacts. The proposed DIREQT method can be combined with motion correction techniques to provide accurate quantitative maps from highly accelerated free-breathing and/or continuously acquired data. Please also read paragraph [0055-0056, 0101, 0124 and 0140-0141]) is trained using a loss function that comprises a MR physics loss term (Fig. 5. Paragraph [0134]-CORREIA discloses the training step consists of pairs of undersampled k-space (or images) and the desired ground-truth TK parameter maps. Then, the reconstruction can be trained in an end-to-end fashion, in which TK parameter maps are reconstructed with the network from undersampled data and compared to the ground-truth. In paragraph [0135]-CORREIA discloses the trained CNN can be used to generate artefact-free TK parameter maps from undersampled FPP-CMR data. The network is trained to learn the mapping between the undersampled k-space data (or image) and TK parameter maps and output an estimate of residual maps. In paragraph [0136]-CORREIA discloses several loss functions can be used to train deep neural networks. A popular choice is the mean squared error between the TK parameter map estimate and ground truth (or residual). The forward physical model loss function between the input data and model generated data (Eq. 5) could also be included).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA of having a system for magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having wherein the two-stage deep learning network is trained using a loss function that comprises a MR physics loss term.
Wherein having WANG’s method having wherein the two-stage deep learning network is trained using a loss function that comprises a MR physics loss term.
The motivation behind the modification would have been to obtain a method that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 11, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG fails to explicitly teach further comprising predicting MR parameter quantification error using the at least one uncertainty map.
However, CORREIA explicitly teaches further comprising predicting MR parameter quantification error (Fig. 5. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data (wherein DIREQT can be combined with motion correction techniques to provide accurate quantitative maps from highly accelerated free-breathing data). In paragraph [0133]-CORREIA discloses Deep learning can be used to reconstruct motion corrected tracer-kinetic maps. A solution to accelerate reconstruction time is deep learning to directly estimate TK parameter maps from undersampled FPP-CMR data (DIREQT-NET). In paragraph [0135]-CORREIA discloses the trained CNN can be used to generate artefact-free TK parameter maps from undersampled FPP-CMR data. The network is trained to learn the mapping between the undersampled k-space data and TK parameter maps and output an estimate of residual maps. Please also read paragraph [0134]) using the at least one uncertainty map (Fig. 5. Paragraph [0140]-CORREIA discloses other deep neural networks than CNNs can be used for DIREQT-NET, such as recurrent neural networks, (cycle) generative adversarial networks, Bayesian neural networks, ADMM-Net, etc. In paragraph [0141]-CORREIA discloses the use of a Bayesian neural network may be beneficial because it may additionally provide uncertainty maps which may be useful in assessing the accuracy of the motion corrected tracer-kinetic map).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA of having a system for magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having further comprising predicting MR parameter quantification error using the at least one uncertainty map.
Wherein having WANG’s method having further comprising predicting MR parameter quantification error using the at least one uncertainty map.
The motivation behind the modification would have been to obtain a method that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 12, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG further teaches wherein the at least one quantitative MR parameter map (Fig. 33. Paragraph [0232]-WANG discloses one or more of the above quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33) includes a proton-density fat fraction (PDFF) map, a R2* map (Fig. 33. Paragraph [0087]-WANG discloses R2* information is essential in the disclosed preconditioning technique. The proposed method incorporates R2* contrast into the construction of preconditioner in a more adaptive manner. After obtaining a rough estimate of tissue susceptibilities), and a Bo field map (Fig. 33. Paragraph [0100]-WANG discloses U-Net, a fully convolutional neural network, was chosen as the example network structure. The network was designed an input/output patch at size 128×128×24, while the original 3D volume field map was segmented into patches using a scheme of 66% overlapping between adjacent patches. The output patches were compiled to recover the full volume).
Regarding claim 13, WANG in view of CORREIA explicitly teach the method according to claim 1, WANG further teaches wherein the quantitative MR parameter is one of Ti, T2, stiffness, susceptibility (Fig. 1. Paragraph [0067]-WANG discloses to fully exploit the capacity of preconditioning in accelerating reconstruction of Quantitative Susceptibility Mapping. The present inventors propose an automated generation of an adaptive preconditioner from the total field f and R2*, as illustrated in FIG. 1. An approximate susceptibility map is estimated rapidly from the field input f.), diffusion (Fig. 1. Paragraph [0056]-WANG discloses implementations of systems and methods for collecting and processing MRI signals of a subject, reconstructing maps of physical properties intrinsic to the subject (e.g., magnetic susceptibility, transport parameters), and reconstructing multiple contrast images are described. Physical processes in tissue including magnetism, transport, and relaxation affect of MRI signals. Image data acquired in a time resolved manner during a contrast agent passage through tissue allow determination of tissue transport properties including diffusion (wherein a neural network is used to estimate transport parameters, such as diffusion). Images of various contrasts including T1, T2, T2* and diffusion weightings have structural consistency. Please see paragraph [0058]), chemical exchange, or magnetization transfer (Fig. 1. Paragraph [0056]-WANG discloses phase data acquired at various echo times in a gradient echo sequence allow determination of the magnetic field generated by tissue susceptibility and therefore determination of susceptibility. In paragraph [0236]-WANG discloses the magnetic field can be determined from the complex MRI data by fitting the detected signal as a sum over tissue spectral components, where each component signal characterized by an exponent with a negative real part representing signal decay and an imaginary part representing phase dependent on the magnetic field and the chemical shift (step 830). Please also read paragraph [0232]).
Regarding claim 15, WANG explicitly teaches a system for magnetic resonance imaging (MRI) quantitative parameter maps (Fig. 34. Abstract-WANG discloses exemplary methods for quantitative mapping of physical properties, systems and computer-accessible medium can be provided to generate images of tissue magnetic susceptibility, transport parameters and oxygen consumption from magnetic resonance imaging data using the Bayesian inference approach. In paragraph [0252]-WANG discloses the described techniques can be performed using a computer system. FIG. 34 is a block diagram of an example computer system 900 that can be used to perform implementations of the process 800. Please also see Fig. 33) comprising:
an input for receiving at least one multi-contrast magnetic resonance (MR) image of a subject (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. Please also read paragraph [0136-0137, 0233, 0248]);
a two-stage deep learning network (Fig. 33. Paragraph [0120]-WANG discloses the fidelity imposed network edit (FINE) is an artificial neural network, including convolutional neural network. FINE can achieve superior performance in quantitative susceptibility mapping (QSM) and undersampled multi-contrast reconstruction in MRI. Please also read paragraph [0136-0137, 0232, and 0237-0239]) comprising:
an artifact suppression deep learning network configured to generate at least one multi-contrast MR image with suppressed undersampling artifacts using the at least one multi-contrast MR image of the subject (Fig. 33. Paragraph [0136]-WANG discloses FINE was applied to multi-contrast MRI reconstruction with under-sampled data. In order to accelerate the time-consuming acquisition of certain contrasts, such as T2 weighted (T2w) or T2 Fluid Attenuated Inversion Recovery (T2FLAIR) images, k-space was under-sampled, thus requiring a regularized algorithm to recover images with minimal artifact. Please also read paragraph [0232]); and
a display coupled to the two-stage deep learning network and configured to display at least one of the at least one multi-contrast MR image with suppressed undersampling artifacts, the at least one quantitative MR parameter map, and the corresponding uncertainty estimation map (Fig. 33. Paragraph [0232]-WANG discloses using these quantitative maps of physical properties and multiple contrasts, one or more images of the subject can be generated and displayed to a user. The user can then use these images for diagnostic, therapeutic or experimental purposes, for example to investigate the structure and/or composition and/or function of the subject, and/or to diagnose various conditions or diseases based, and/or to treat various conditions or diseases based, at least in part, on the images. Please also read paragraph [0240, 0247 and 0251]).
Although WANG explicitly teaches a parameter mapping deep learning network coupled to the artifact suppression deep learning network, the parameter mapping deep learning network configured to generate at least one quantitative MR parameter map based on the at least one multi-contrast MR image with suppressed undersampling artifacts; and (Fig. 33. Paragraph [0232]-WANG discloses the process 800 can be used to map the tissue magnetic susceptibility of a subject. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. The process 800 can be used to obtain time-resolved MR data corresponding a subject as a contrast agent passing through tissue in an organ of the subject, and process this time-resolved MR data to generate a quantitative transport map of the subject (wherein a neural network may be used to perform each of these tasks). Please also read [0136, 0237-0239, 0245-0246, and 0249-0250]).
WANG fails to explicitly teach the parameter mapping deep learning network configured to generate an uncertainty estimation map for the at least one quantitative MR parameter map.
However, CORREIA explicitly teaches the parameter mapping deep learning network (Fig. 5. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data (wherein DIREQT can be combined with motion correction techniques to provide accurate quantitative maps from highly accelerated free-breathing data). In paragraph [0133]-CORREIA discloses Deep learning can be used to reconstruct motion corrected tracer-kinetic maps. A solution to accelerate reconstruction time is deep learning to directly estimate TK parameter maps from undersampled FPP-CMR data (DIREQT-NET). In paragraph [0135]-CORREIA discloses the trained CNN can be used to generate artefact-free TK parameter maps from undersampled FPP-CMR data. The network is trained to learn the mapping between the undersampled k-space data and TK parameter maps and output an estimate of residual maps. Please also read paragraph [0134]) configured to generate an uncertainty estimation map for the at least one MR parameter map (Fig. 5. Paragraph [0140]-CORREIA discloses other deep neural networks than CNNs can be used for DIREQT-NET, such as a Bayesian neural networks. In paragraph [0141]-CORREIA discloses the use of a Bayesian neural network may additionally provide uncertainty maps which may be useful in assessing the accuracy of the motion corrected tracer-kinetic map); and
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG having a system for magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having the parameter mapping deep learning network configured to generate an uncertainty estimation map for the at least one quantitative MR parameter map.
Wherein having WANG’s method having a parameter mapping deep learning network coupled to the artifact suppression deep learning network, the parameter mapping deep learning network configured to generate at least one quantitative MR parameter map based on the at least one multi-contrast MR image with suppressed undersampling artifacts and to generate an uncertainty estimation map for the at least one quantitative MR parameter map; and
The motivation behind the modification would have been to obtain a method that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 16, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG further teaches wherein one or more of the at least one multi-contrast MR image and at least one multi-contrast MR image with suppressed undersampling artifacts are multi-echo MR images (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. Please also read paragraph [0136]).
Regarding claim 17, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG further teaches further comprising a pre-processing module coupled to the two-stage deep learning network and configured to generate the at least one multi-contrast MR image of the subject from undersampled k-space data (Fig. 33. Paragraph [0136]-WANG discloses FINE was applied to multi-contrast MRI reconstruction with under-sampled data. In order to accelerate the time-consuming acquisition of certain contrasts, such as T2 weighted (T2w) or T2 Fluid Attenuated Inversion Recovery (T2FLAIR) images, k-space was under-sampled, thus requiring a regularized algorithm to recover images with minimal artifact. In paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to map the tissue magnetic susceptibility of a subject, such as a patient (or portion of a patient). The process 800 can be used to transform magnetic resonance (MR) signal data corresponding to a subject into multiple images that quantitatively depict the structure and/or composition and/or function of the subject. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images).
Regarding claim 19, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG fails to explicitly teach further comprising a post-processing module coupled to the two-stage deep learning network and configured to predict MR parameter quantification error using the at least one uncertainty map.
However, CORREIA explicitly teaches further comprising a post-processing module coupled to the two-stage deep learning network (Fig. 4. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data. In paragraph [0133]-CORREIA discloses deep learning can be used to reconstruct motion corrected tracer-kinetic maps. A potential solution to accelerate reconstruction time is to use a deep learning approach to directly estimate TK parameter maps from undersampled FPP-CMR data (DIREQT-NET). In paragraph [0134]-CORREIA discloses one approach to solve Eq. (5) using deep learning is to directly learn the nonlinear mapping between the undersampled k-space data d or aliased zero-filled undersampled reconstruction and the fully-sampled TK parameter maps using a deep neural network (e.g. convolutional neural network, CNN). In paragraph [0135]-CORREIA discloses the trained CNN can be used to generate artefact-free TK parameter maps from undersampled FPP-CMR data. The network is trained to learn the mapping between the undersampled k-space data (or image) and TK parameter maps and output an estimate of residual maps) and configured to predict MR parameter quantification error using the at least one uncertainty map (Fig. 4. Paragraph [0140]-CORREIA discloses other deep neural networks than CNNs can be used for DIREQT-NET, such as a Bayesian neural networks. In paragraph [0141]-CORREIA discloses the use of a Bayesian neural network may additionally provide uncertainty maps which may be useful in assessing the accuracy of the motion corrected tracer-kinetic map).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA having a system for magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having further comprising a post-processing module coupled to the two-stage deep learning network and configured to predict MR parameter quantification error using the at least one uncertainty map.
Wherein having WANG’s system having further comprising a post-processing module coupled to the two-stage deep learning network and configured to predict MR parameter quantification error using the at least one uncertainty map.
The motivation behind the modification would have been to obtain a system that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 20, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG further teaches wherein the artifact suppression learning network is a convolutional neural network (Fig. 33. Paragraph [0120]-WANG discloses described is an example fidelity imposed network edit (FINE) where the network is an artificial neural network in deep learning, including a convolutional neural network. The experiments demonstrated that FINE can achieve superior performance in quantitative susceptibility mapping (QSM) and undersampled multi-contrast reconstruction in MRI. Please also read paragraph [0132, 0138, 0187 and 0232]).
Regarding claim 21, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG further teaches wherein the parameter mapping deep learning network is a convolutional neural network (Fig. 33. Paragraph [0120]-WANG discloses described is an example fidelity imposed network edit (FINE) where the network is an artificial neural network in deep learning, including a convolutional neural network. The experiments demonstrated that FINE can achieve superior performance in quantitative susceptibility mapping (QSM) and undersampled multi-contrast reconstruction in MRI. Please also read paragraph [0132, 0138, 0187 and 0232]).
Regarding claim 22, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG fails to explicitly teach wherein the two-stage deep learning network is trained using a loss function that comprises a MR physics loss term.
However, CORREIA explicitly teaches wherein the two-stage deep learning network (Fig. 5. Paragraph [0099]-CORREIA discloses a DIRect QuanTitative (DIREQT) FPP-CMR reconstruction framework is proposed to directly estimate quantitative myocardial perfusion maps from undersampled data (wherein DIREQT-NET may be convolutional neural networks or Bayesian neural networks with uncertainty maps). In paragraph [0131]-CORREIA discloses in order to obtain accurate quantitative maps, the first-pass data must be motion compensated to minimise for the resulting artefacts. The proposed DIREQT method can be combined with motion correction techniques to provide accurate quantitative maps from highly accelerated free-breathing and/or continuously acquired data. Please also read paragraph [0055-0056, 0101, 0124 and 0140-0141]) is trained using a loss function that comprises a MR physics loss term (Fig. 5. Paragraph [0134]-CORREIA discloses the training step consists of pairs of undersampled k-space (or images) and the desired ground-truth TK parameter maps. Then, the reconstruction can be trained in an end-to-end fashion, in which TK parameter maps are reconstructed with the network from undersampled data and compared to the ground-truth. In paragraph [0135]-CORREIA discloses the trained CNN can be used to generate artefact-free TK parameter maps from undersampled FPP-CMR data. The network is trained to learn the mapping between the undersampled k-space data (or image) and TK parameter maps and output an estimate of residual maps. In paragraph [0136]-CORREIA discloses several loss functions can be used to train deep neural networks. A popular choice is the mean squared error between the TK parameter map estimate and ground truth (or residual). The forward physical model loss function between the input data and model generated data (Eq. 5) could also be included).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA having a system for magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of CORREIA having wherein the two-stage deep learning network is trained using a loss function that comprises a MR physics loss term.
Wherein having WANG’s system having wherein the two-stage deep learning network is trained using a loss function that comprises a MR physics loss term.
The motivation behind the modification would have been to obtain a system that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and CORREIA systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while CORREIA provides systems and methods that can be combined with respiratory motion correction and k-t undersampling to improve myocardial perfusion quantification, while substantially increasing patient comfort, greatly reducing the amount of data necessary to obtain high-quality TK parameter maps and improving spatial and temporal resolutions. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and CORREIA et al. (US 20220373631 A1), Abstract and Paragraph [0096, 0124, 0130].
Regarding claim 23, WANG in view of CORREIA explicitly teach the system according to claim 15, WANG further teaches wherein the at least one quantitative MR parameter map (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33) includes a proton-density fat fraction (PDFF) map, a R2* map (Fig. 33. Paragraph [0087]-WANG discloses R2* information is essential in the disclosed preconditioning technique. The proposed method incorporates R2* contrast into the construction of preconditioner in a more adaptive manner. After obtaining a rough estimate of tissue susceptibilities), and a Bo field map (Fig. 33. Paragraph [0100]-WANG discloses U-Net, a fully convolutional neural network, was chosen as the example network structure. The network was designed an input/output patch at size 128×128×24, while the original 3D volume field map was segmented into patches using a scheme of 66% overlapping between adjacent patches. The output patches were compiled to recover the full volume).
Regarding claim 24, WANG in view of CORREIA explicitly teach he system according to claim 15, WANG further teaches wherein the quantitative MR parameter (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33) is one of Ti, T2, stiffness, susceptibility (Fig. 1. Paragraph [0067]-WANG discloses to fully exploit the capacity of preconditioning in accelerating reconstruction of Quantitative Susceptibility Mapping. The present inventors propose an automated generation of an adaptive preconditioner from the total field f and R2*, as illustrated in FIG. 1. An approximate susceptibility map is estimated rapidly from the field input f.), diffusion (Fig. 1. Paragraph [0056]-WANG discloses implementations of systems and methods for collecting and processing MRI signals of a subject, reconstructing maps of physical properties intrinsic to the subject (e.g., magnetic susceptibility, transport parameters), and reconstructing multiple contrast images are described. Physical processes in tissue including magnetism, transport, and relaxation affect of MRI signals. Image data acquired in a time resolved manner during a contrast agent passage through tissue allow determination of tissue transport properties including diffusion (wherein a neural network is used to estimate transport parameters, such as diffusion). Images of various contrasts including T1, T2, T2* and diffusion weightings have structural consistency. Please see paragraph [0058]), chemical exchange, or magnetization transfer (Fig. 1. Paragraph [0056]-WANG discloses phase data acquired at various echo times in a gradient echo sequence allow determination of the magnetic field generated by tissue susceptibility and therefore determination of susceptibility. In paragraph [0236]-WANG discloses the magnetic field can be determined from the complex MRI data by fitting the detected signal as a sum over tissue spectral components, where each component signal characterized by an exponent with a negative real part representing signal decay and an imaginary part representing phase dependent on the magnetic field and the chemical shift (step 830). Please also read paragraph [0232]).
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over WANG et al. (US 20220229140 A1), hereinafter referenced as WANG in view of CORREIA et al. (US 20220373631 A1), hereinafter referenced as CORREIA and in further view of POLAK et al. (US 20210264645 A1), hereinafter referenced as POLAK.
Regarding claim 6, WANG in view of CORREIA explicitly teach the method according to claim 5, WANG in view of CORREIA fails to explicitly teach wherein the plurality of multi-contrast MR images are stacked along the channel dimension.
However, POLAK explicitly teaches wherein the plurality of multi-contrast MR images are stacked along the channel dimension (Fig. 2. Paragraph [0118]-POLAK discloses a multi-contrast (MC) reconstruction procedure may use Deep Learning to jointly reconstruct multiple highly under-sampled MRI contrasts, such as T1w, T2w, T2*w and the like, acquired using complementary k-space under-sampling. In paragraph [0121]-POLAK discloses multi-contrast reconstructions expand the network by stacking multiple clinical contrasts along the channel dimension of the network).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA having a method for generating magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of POLAK having wherein the plurality of multi-contrast MR images are stacked along the channel dimension.
Wherein having WANG’s method having wherein the plurality of multi-contrast MR images are stacked along the channel dimension.
The motivation behind the modification would have been to obtain a method that improves image quality and accuracy for multi-contrast reconstruction, since both WANG and POLAK concern systems and methods for MRI, deep learning and multi-contrast reconstruction. Wherein WANG provides systems and methods that improves the accuracy in deep learning (DL) solutions of image reconstruction from noisy incomplete data, including ill-posed inverse problems of quantitative susceptibility mapping, quantitative tissue mapping, and multiple contrast image reconstruction, while POLAK provides systems and methods that improve the image quality of Multi-Contrast (MC) Deep Learning (DL) reconstruction methods using undersampled data. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004] and POLAK et al. (US 20210264645 A1), Abstract and Paragraph [0019, 0036, 0097, 0121].
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over WANG et al. (US 20220229140 A1), hereinafter referenced as WANG in view of CORREIA et al. (US 20220373631 A1), hereinafter referenced as CORREIA and in further view of NICKEL (Nickel, Marcel Dominik, Machine Translation of German Patent Publication DE 102020210775 A1 “Magnetic resonance imaging reconstruction using machine learning and motion compensation”, Filed 2020-08-26, Published 2022-03-03), hereinafter referenced as NICKEL.
Regarding claim 8, WANG in view of CORREIA explicitly teach the method according to claim 7, although WANG explicitly teaches the plurality of quantitative MR parameter maps (Fig. 33. Paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to map the tissue magnetic susceptibility of a subject. The process 800 can be used to transform magnetic resonance (MR) signal data corresponding to a subject into multiple images that quantitatively depict the structure and/or composition and/or function of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. The process 800 can be used to obtain time-resolved MR data corresponding a subject as a contrast agent passing through tissue in an organ of the subject, and process this time-resolved MR data to generate a quantitative transport map of the subject. Using these quantitative maps of physical properties and multiple contrasts, one or more images of the subject can be generated and displayed to a user. Please also read paragraph [0237-0239, 0244-0246 and 0248-0250]).
WANG in view of CORREIA fails to explicitly teach wherein each quantitative MR parameter map in the plurality of quantitative MR parameter maps is stacked along the channel dimension.
However, NICKEL explicitly teaches wherein each MR parameter map in the plurality of MR parameter maps is stacked along the channel dimension (Fig. 2. Paragraph [0047]-NICKEL discloses the techniques rely on a common regularization operation that is applied to multiple MRI images obtained in one iteration of iterative optimization. This means that the regularization operation accepts multiple MRI images as one input. The regularization operation has an input that includes a concatenation of several previous mappings obtained from a previous iteration of the multiple iterations. The concatenation can correspond to stacking several previous mappings along a channel dimension of the ML algorithm, for example a deep neural network, where the regularization operation is implemented).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA having a method for generating magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of NICKEL of having wherein the plurality of multi-contrast MR images are stacked along the channel dimension.
Wherein having WANG’s method having wherein each quantitative MR parameter map in the plurality of quantitative MR parameter maps is stacked along the channel dimension.
The motivation behind the modification would have been to obtain a method that improves the quality and accuracy for MRI images and reconstruction, since both WANG and NICKEL concern systems and methods for MRI, deep learning and reconstruction. Wherein WANG provides systems and methods that improves the accuracy in deep learning (DL) solutions of image reconstruction from noisy incomplete data, including ill-posed inverse problems of quantitative susceptibility mapping, quantitative tissue mapping, and multiple contrast image reconstruction, while NICKEL provides systems and methods that adjust deformation operators over the course of iterative optimization based on previous mappings, improves the current deformation operators through each iteration and improves the quality of the current MRI images. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004] and NICKEL (Nickel, Marcel Dominik, Machine Translation of German Patent Publication DE 102020210775 A1 “Magnetic resonance imaging reconstruction using machine learning and motion compensation”, Filed 2020-08-26, Published 2022-03-03), Abstract and Paragraph [0052 and 0057].
Claims 14 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over WANG et al. (US 20220229140 A1), hereinafter referenced as WANG in view of CORREIA et al. (US 20220373631 A1), hereinafter referenced as CORREIA and in further view of ZHONG et al. (US 20210349166 A1), hereinafter referenced as ZHONG.
Regarding claim 14, WANG in view of CORREIA explicitly teach the method according to claim 1, although WANG explicitly teaches wherein the at least one multi-contrast MR image is acquired using an undersampled multi-echo MRI acquisition (Fig. 33. Paragraph [0136]-WANG discloses FINE was applied to multi-contrast MRI reconstruction with under-sampled data. In order to accelerate the time-consuming acquisition of certain contrasts, such as T2 weighted (T2w) or T2 Fluid Attenuated Inversion Recovery (T2FLAIR) images, k-space was under-sampled, thus requiring a regularized algorithm to recover images with minimal artifact (wherein FINE stands for fidelity imposed network edit, which is an artificial neural network, including convolutional neural network). Further in paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. Please also read paragraph [0136-0137]).
WANG in view of CORREIA fail to explicitly teach wherein the at least one multi-contrast MR image is acquired using an undersampled free-breathing multi-echo stack-of-radial MRI acquisition.
However, ZHONG explicitly teaches wherein the at least one multi-contrast MR image (Fig. 1. Paragraph [0007]-ZHONG discloses a method of generating biomarker parameters includes acquiring imaging data depicting a patient using a MRI system. The imaging data is acquired for a plurality of contrasts resulting from application of a pulse on the patient's anatomy. A process is executed to generate a motion correction and average (MoCoAve) image for each contrast. This process includes dividing the imaging data for the contrast into a plurality of bin corresponding to one of a plurality of respiratory motion phases, and reconstructing the imaging data in each bin to yield a plurality of bin image. The process further includes selecting a reference bin image from the plurality of bin images, and warping the plurality of bin images. The warped bin images and the reference bin image are averaged to generate the MoCoAve image for the contrast. One or more biomarker parameter maps are calculated based on the MoCoAve images generated for the plurality of contrasts) is acquired using free-breathing multi-echo stack-of-radial MRI acquisition (Fig. 1. Paragraph [0018]-ZHONG discloses FIGS. 1A-1B show a flow chart depicting the data acquisition and data processing steps for the proposed free-breathing MRI method utilizing stack-of-radial imaging acquisition with self-gating and MoCoAve and multi-echo Dixon techniques. In paragraph [0025]-ZHONG discloses at Step 140A the multi-echo MoCoAve images reconstructed in Step 135A or 135B are used to calculate the subsequent quantitative biomarker parameters such as PDFF, R2* and LIC. In paragraph [0030]-ZHONG discloses as shown in FIG. 2A, a self-gating signal is retrospectively extracted in a single continuous free-breathing scan acquired with a stack-of-radial imaging acquisition. The data points of the radial readout views at the k-space origin are sampled as the corresponding self-gating signal).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA of having a method for generating magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of ZHONG having wherein the at least one multi-contrast MR image is acquired using free-breathing multi-echo stack-of-radial MRI acquisition.
Wherein having WANG’s method having wherein the at least one multi-contrast MR image is acquired using an undersampled free-breathing multi-echo stack-of-radial MRI acquisition.
The motivation behind the modification would have been to obtain a method that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and ZHONG systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while ZHONG provides systems and methods that improve the quality and signal-to-noise ratio of free-breathing quantitative measurement of magnetic resonance imaging (MRI) parameters and related biomarkers. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and ZHONG et al. (US 20210349166 A1), Abstract and Paragraph [0001, 0005-0006, and 0017-0019].
Regarding claim 18, WANG in view of CORREIA explicitly teach the system according to claim 17, although WANG explicitly teaches wherein the undersampled k-space data is acquired using a multi-echo MRI acquisition (Fig. 33. Paragraph [0136]-WANG discloses FINE was applied to multi-contrast MRI reconstruction with under-sampled data. In order to accelerate the time-consuming acquisition of certain contrasts, such as T2 weighted (T2w) or T2 Fluid Attenuated Inversion Recovery (T2FLAIR) images, k-space was under-sampled, thus requiring a regularized algorithm to recover images with minimal artifact (wherein FINE stands for fidelity imposed network edit, which is an artificial neural network, including convolutional neural network). Further in paragraph [0232]-WANG discloses one or more of the quantitative mapping techniques can be implemented using the process 800 shown in FIG. 33. The process 800 can be used to obtain multiecho MR data corresponding to a subject, and process this multiecho MR data to generate a quantitative susceptibility map of the subject. The process 800 can be used to obtain rapidly undersampled multiple contrast MR data corresponding to a subject, and process this data to generate multiple contrast MR images. Please also read paragraph [0136-0137]).
WANG in view of CORREIA fail to explicitly teach wherein the undersampled k-space data is acquired using a self-gating free-breathing multi-echo stack-of-radial MRI acquisition.
However, ZHONG explicitly teaches wherein the k-space data is acquired using a self-gating free-breathing multi-echo stack-of-radial MRI acquisition (Fig. 1. Paragraph [0018]-ZHONG discloses FIGS. 1A-1B show a flow chart depicting the data acquisition and data processing steps for the proposed free-breathing MRI method utilizing stack-of-radial imaging acquisition with self-gating and MoCoAve and multi-echo Dixon techniques. In paragraph [0025]-ZHONG discloses at Step 140A the multi-echo MoCoAve images reconstructed in Step 135A or 135B are used to calculate the subsequent quantitative biomarker parameters such as PDFF, R2* and LIC. In paragraph [0030]-ZHONG discloses as shown in FIG. 2A, a self-gating signal is retrospectively extracted in a single continuous free-breathing scan acquired with a stack-of-radial imaging acquisition. The data points of the radial readout views at the k-space origin are sampled as the corresponding self-gating signal).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of WANG in view of CORREIA of having a system for magnetic resonance imaging (MRI) quantitative parameter maps, with the teachings of ZHONG having wherein the k-space data is acquired using a self-gating free-breathing multi-echo stack-of-radial MRI acquisition.
Wherein having WANG’s system having wherein the k-space data is acquired using a self-gating free-breathing multi-echo stack-of-radial MRI acquisition.
The motivation behind the modification would have been to obtain a system that improves medical diagnostics and the accuracy and quality of MRI images, since both WANG and ZHONG systems and methods for MRI images, deep learning and quantitative mapping. Wherein WANG provides systems and methods that overcome limitations in quantitative susceptibility mapping (QSM), where the computation speed and accuracy and the Bayesian prior information affect QSM image quality and practical usage, reduces shadow and streaking artifacts, improves the quality and/or accuracy of susceptibility and/or transport-based images, improves a user's understanding of a subject's structure and/or composition and/or function, and improves the accuracy of any resulting medical diagnoses or therapeutic analyses, while ZHONG provides systems and methods that improve the quality and signal-to-noise ratio of free-breathing quantitative measurement of magnetic resonance imaging (MRI) parameters and related biomarkers. Please see WANG et al. (US 20220229140 A1), Abstract and Paragraph [0004, 0057, 0061, 0232] and ZHONG et al. (US 20210349166 A1), Abstract and Paragraph [0001, 0005-0006, and 0017-0019].
Conclusion
Listed below are the prior arts made of record and not relied upon but are considered pertinent to applicant`s disclosure.
CHEN et al. (US 20220308147 A1)- Systems and methods providing enhancements to quantitative imaging systems and techniques are described herein. In one aspect, a system for tissue quantification in magnetic resonance fingerprinting (MRF) comprises a feature extraction module operable to convert pixel input high-dimensional signal evolution in to a low-dimensional feature map. The system also comprises a spatially constrained quantification module operable to capture spatial information from the low-dimensional feature map and generate an estimated tissue property map.............................. Please see Fig. 1-3. Abstract.
Ozdemir et al. (US 20190122073 A1)- This invention provides a system and method to propagate uncertainty information in a deep learning pipeline. It allows for the propagation of uncertainty information from one deep learning model to the next by fusing model uncertainty with the original imagery dataset. This approach results in a deep learning architecture where the output of the system contains not only the prediction, but also the model uncertainty information associated with that prediction. The embodiments herein improve upon existing deep learning-based models (CADe models) by providing the model with uncertainty/confidence information associated with (e.g. CADe) decisions. This uncertainty information can be employed in various ways, including (a) transmitting uncertainty from a first stage (or subsystem) of the machine learning system into a next (second) stage (or the next subsystem), and (b) providing uncertainty information to the end user in a manner that characterizes the uncertainty of the overall machine learning model............................ Please see Fig. 1-5. Para. [0031, 0035 and 0038]. Abstract.
YOO et al. (US 20200302596 A1)- Methods and systems are provided for automatically estimating image-level uncertainty for MS lesion segmentation data. A segmentation network is trained to segment MS lesions. The trained segmentation network is then used to estimate voxel level measures of uncertainty by performing Monte-Carlo (MC) dropout. The estimated voxel level uncertainty measures are converted into lesion level uncertainty measures. The information density of the lesion mask, the voxel level measures, and the lesion level measures is increased. A trained network receives input images, the segmented lesion masks, the voxel level uncertainty measures, and the lesion level uncertainty measures and outputs an image level uncertainty measure. The network is trained with a segmentation performance metric to predict an image level uncertainty measure on the segmented lesion mask that is produced by the trained segmentation network............................. Please see Fig. 1-6. Abstract.
NOVIKOV et al. (US 20210076972 A1)- An exemplary system, method and computer-accessible medium for generating a denoised magnetic resonance (MR) image(s) of a portion(s) of a patient(s) can be provided, which can include, for example, generating a plurality of MR images of the portion(s), where a number of the MR images can be based on a number of MR coils in a MR apparatus used to generate the MR images, generating MR imaging information by denoising a first one of the MR images based on another one of the MR images, and generating the denoised MR image(s) based on the MR imaging information. The number of the MR coils can be a subset of a total number of the MR coils in the MR apparatus. The number of the MR coils can be a total number of the MR coils in the MR apparatus. The MR information can be generated by denoising each of the MR images based on the other one of the MR images.............................. Please see Par. [0047, 0056, 0079-0085, 0100, 0102]. Abstract.
Hawkins-Daarud et al. (US 20220148731 A1)- Genetic and/or other biological marker prediction data are generated based on inputting medical image data to a suitably trained machine learning model, where the output genetic prediction data not only indicate a prediction of genetic features and/or other biological markers for a subject, but also a measure of uncertainty in each of those predictions. As an example, a transductive learning Gaussian process model is used to generate the genetic and/or other biological marker predication data and corresponding predictive uncertainty data. As another example, a knowledge-infused global-local data fusion model can be used for spatial predictive modeling............................... Please see Fig. 1-3 and 7. Abstract.
CAI et al. (US 20190302211 A1)- In some aspects, the disclosed technology relates to free-breathing cine DENSE (displacement encoding with stimulated echoes) imaging. In some embodiments, self-gated free-breathing adaptive acquisition reduces free-breathing artifacts by minimizing the residual energy of the phase-cycled T1-relaxation signal, and the acquisition of the k-space data is adaptively repeated with the highest residual T1-echo energy. In some embodiments, phase-cycled spiral interleaves are identified at matched respiratory phases by minimizing the residual signal due to T1 relaxation after phase-cycling subtraction; image-based navigators (iNAVs) are reconstructed from matched phase-cycled interleaves that are comprised of the stimulated echo iNAVs (ste-iNAVs), wherein the ste-iNAVs are used for motion estimation and compensation of k-space data............................. Please see Fig. 2-3. Abstract.
WANG et al. (US 20110044524 A1)- A method and apparatus is provided for magnetic source magnetic resonance imaging. The method includes collecting energy signals from an object, providing additional information of characteristics of the object, and generating the image of the object from the energy signals and from the additional information such that the image includes a representation of a quantitative estimation of the characteristics, e.g a quantitative estimation of magnetic susceptibility. The additional information may comprise predetermined characteristics of the object, a magnitude image generated from the object, or magnetic signals collected from different relative orientations between the object and the imaging system. The image is generated by an inversion operation based on the collected signals and the additional information. The inversion operation minimizes a cost function obtained by combining the data extracted from the collected signals and the additional information of the object. Additionally, the image is used to detect a number of diagnostic features including microbleeds, contract agents and the like.............................. Please see Fig. 1. Para. [0454, 0827, 1021]. Abstract.
Any inquiry concerning this communication or earlier communications from the examiner
should be directed to Aaron Bonansinga whose telephone number is (703) 756-5380 The examiner can normally be reached on Monday-Friday, 9:00 a.m. - 6:00 p.m. ET.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s
supervisor, Chineyere Wills-Burns can be reached by phone at (571) 272-9752. 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.
/AARON TIMOTHY BONANSINGA/Examiner, Art Unit 2673
/CHINEYERE WILLS-BURNS/Supervisory Patent Examiner, Art Unit 2673