DETAILED ACTION This action is in response to the filing of 2-12-2024 . Claims 1- 20 are pending and have been considered below: Claim Objections Claim 20 is objected to because of the following informalities: Claim 20 recites an “apparatus” of claim 1, however claim 1 is a method. Appropriate correction is required. 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, 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- 6, 8 and 16-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over “ Gaussian Prototypical Networks for Few-Shot Learning on Omniglot ”; Stanislav For t(“Fort”) ; pages 1-4, 2017 in view of “ Training Strategies for Radiology Deep Learning Models in Data-limited Scenarios ”, Sema Candemir et al. (“Candemir”) , Pages 1-10, 2021 and “Meta-learning with implicit gradients in a few-shot setting for medical image segmentation”, Khadka et al. (“Khadha”) , pages 1-10, 1-12-2022 . Claim 1 : Fort discloses a method comprising: performing meta-learning for a variance-aware prototypical machine learning network pre-trained on a dataset comprising examples of a first type (Fort: abstract and Introduction; Gaussian prototypical network) , wherein: Fort may not explicitly disclose a dataset of radiology report associated with a single domain . Candemir is provided to disclose a radiology dataset domain (Pages 2-3: Transfer learning with a same modality dataset; Paragraph 2 and Figure 1; provides radiology domain (i.e. MRI images)). Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate radiology domains with the dataset provided by Fort . One would have been motivated to provide the functionality because the domain specific image provides improved training pertaining to the specificity of medical imaging . Fort also may not explicitly disclose the meta-learning comprises learning one or more respective prototype representations for each radiology classification task included in the meta-learning and a respective variance information for the one or more respective prototype representations for each radiology classification task included in the meta-learning ; Khadha is provided because it disclose s meta-learning for medical imaging (abstract). Khadha further discloses meta-learning for classification of a representation (Pages 2 , Column 1, Paragraph 3 “ meta-learning under few-shot settings has emerged as a potential solution [19,20], especially in limited data settings. Meta-learning enables learning model weights by leveraging prior knowledge from various tasks [21] and can be implemented in different task objectives such as few-shot learning or multi-task learning. It is advantageous to use meta-learning in few-shot settings, and it has been primarily used in image classification [22,23]. Few-shot learning is a method that uses few annotated examples (support set) to make predictions on unlabeled examples (query set) and is the most appropriate choice when only limited data samples are available. An episodic training in a meta-learning setting can exploit to generalize to such limited data settings and become a natural choice for other tasks such as segmentation. Few-shot learning for segmentation has mostly been explored for natural images [24,25]. Recently, it is gaining more attention in the medical image segmentation [11,26–30]. Recent work by Ref. [11] used a semi-supervised few-shot learning approach to perform skin lesion segmentation by feeding the learner with unlabeled surrogate tasks [31]. applied a few-shot technique with a squeeze and excite block architecture to perform volumetric segmentation of multiple organs in medical images. ” , Figure 1 and Page 5: Section 4.2.3 ). Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate the meta-learning classification into Fort . One would have been motivated to provide the functionality because meta-learning provides improved training for a scarcity of datasets . Khadha further discloses each radiology classification task included in the meta-learning is associated with a corresponding type of radiology report different from the first type of radiology report used for pre-training ; included in the meta-learning (Khadha : Figure 1: Stage 2; support set CVC-612 different from unseen dataset ) ; Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate different datasets for meta-learning f or training in Fort . One would have been motivated to provide the different dataset functionality because it can improve the meta-learning and limit overfitting . modeling the one or more respective prototype representations for each classification task as a Gaussian; providing, to the variance-aware prototypical machine learning network (Fort: abstract and Introduction; Gaussian prototypical network) , , a query sample comprising text data of a type of radiology report seen during the meta-learning ( Fort: Page 2; Section(2), Paragraph 2; query example and Khadka: Figure 1 and Page 3, Section 3.1; query set) ; determining a distance metric between a Dirac distribution representation of the query sample and the Gaussians of the respective prototype representations for each classification task included in the meta-learning ( Fort: Page 2; Section(2), Paragraph 2 and Figure 1; distance metric and Khadha: Pages 2, Column 1, Paragraph 3 ; meta-learning ) ; and classifying the query sample based on identifying a respective prototype representation having the smallest distance metric (Fort: Page 2; Section(2), Paragraph 2 and Figure 1; classify from distance metric) ; Claim 2 : Fort , Candemir and Khadha disclose a method of claim 1, wherein determining the distance metric is based on determining a Wasserstein-Bures metric between Gaussians (Khadha : page 2, Paragraph 3 ; Wasserstein-Bures ) . Claim 3 : Fort, Candemir and Khadha disclose a method of claim 1, wherein the respective variance information for a meta-learning radiology classification task is determined as a covariance matrix over the respective prototype representations for the meta-learning radiology classification task (Fort: Pages 1-2 M ethods ; covariance matrix, Figure 1; classification and Khadha: Pages 2, Column 1, Paragraph 3 ; meta-learning ) . Claim 4 : Fort, Candemir and Khadha disclose a method of claim 3, wherein the distance metric is determined based on the respective variance information and a mean embedding over a plurality of support sample embeddings seen during few-shot learning applied to the variance-aware prototypical network for the meta-learning radiology classification task ( Fort: Pages 1-2 Methods; covariance matrix, Figure 1; distance and Khadka: abstract , Page 4 , Paragraph 1, medical classification ) . Claim 5 : Fort, Candemir and Khadha disclose a method of claim 1, wherein each respective prototype representation used by the variance-aware prototypical network comprises a conditional distribution represented using a Gaussian with a diagonal covariance matrix (Fort: Pages 1-2 methods (b) Diagonal) . Claim 6 : Fort, Candemir and Khadha disclose a method of claim 5, wherein modeling the one or more respective prototype representations as a Gaussian comprises determining the diagonal covariance matrix (Fort: Pages 1-2 methods (b) Diagonal) . Claim 8: Fort, Candemir and Khadha disclose an method of claim 1, further comprising regularizing the variance-aware prototypical machine learning network using a modified loss function to decrease a clustering distance from cluster centroids (Fort: abstract, methods (b) Page 2, Paragraph 4 and Figure 1; cluster distance) . Claim 16 is similar in scope to claim 1 and therefore rejected under the same rationale. Claim 17 is similar in scope to claim 2 and therefore rejected under the same rationale. Claim 18 is similar in scope to claim 3 and therefore rejected under the same rationale. Claim 19 is similar in scope to claim 4 and therefore rejected under the same rationale. Claim 20 : Fort, Candemir and Khadha disclose a n apparatus of claim 1, wherein: each respective prototype representation used by the variance-aware prototypical network comprises a conditional distribution represented using a Gaussian with a diagonal covariance matrix; and modeling the one or more respective prototype representations as a Gaussian comprises determining the diagonal covariance matrix (Fort: Pages 1-2 methods (b) Diagonal) . Claim 7 is/are rejected under 35 U.S.C. 103 as being unpatentable over “ Gaussian Prototypical Networks for Few-Shot Learning on Omniglot ”; Stanislav For t(“Fort”) , pages 1-4, 2017 , “ Training Strategies for Radiology Deep Learning Models in Data-limited Scenarios ”, Sema Candemir et al. (“Candemir”) , Pages 1-10, 2021 and “Meta-learning with implicit gradients in a few- shot setting for medical image segmentation”, Khadka et al. (“Khadha”), pages 1-10, 1-12-2022 in further view of Fujimoto et al. (“Fujimoto” 20230377374 A1). Claim 7 : Fort, Candemir and Khadha disclose a method of claim 1, wherein the covariance information comprises a covariance matrix, and wherein modeling the one or more respective prototype representations ( Fort: Pages 1-2 methods (b) Diagonal) ; however may not explicitly disclose as a Gaussian comprises determining an isotropic Gaussian by averaging over diagonal entries of the covariance matrix . Fujimoto is provided because it discloses an isotropic gaussian with a covariance matrix (Paragraph 51 ). Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate an isotropic gaussian with the diagonal matrix of the modified Fort . One would have been motivated to provide the functionality because it provides enhanced analysis for improved evaluation . Claims 9-10 is/are rejected under 35 U.S.C. 103 as being unpatentable over “ Gaussian Prototypical Networks for Few-Shot Learning on Omniglot ”; Stanislav For t(“Fort”); pages 1-4, 2017 , Training Strategies for Radiology Deep Learning Models in Data-limited Scenarios Sema Candemir et al. (“Candemir”) Pages 1-10, 2021 and “Meta-learning with implicit gradients in a few-shot setting for medical image segmentation”, Khadka et al. (“Khadha”), pages 1-10, 1-12-2022 in further view of Vivona et al. (“Vivona” 20220129706 A1). Claim 9 : Fort, Candemir and Khadha disclose a method of claim 8, but may not explicitly disclose wherein the modified loss function comprises a negative log likelihood loss including a Frobenius norm of a covariance matrix corresponding to the covariance information (Fort: Page 1-2: Methods ; covariance ) . Vivona is provided because it discloses a learning functionality (abstract) and further utilizes loss determination of covariance matrix through Frobenius norm (Paragraph 96). Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate Frobenius norm with the covariance matrix of Fort . One would have been motivated to provide the functionality because it provides enhanced analysis for improved evaluation. Claim 10 : Fort, Candemir , Khadha and Vivona disclose a method of claim 9, wherein the Frobenius norm of the covariance matrix is averaged over all classes in a given meta-batch (Fort: Page 1-2: Methods and Vivona: Paragraph 96; average) . Claim 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over “ Gaussian Prototypical Networks for Few-Shot Learning on Omniglot ”; Stanislav For t(“Fort”); pages 1-4, 2017 , “ Training Strategies for Radiology Deep Learning Models in Data-limited Scenarios ” Sema Candemir et al. (“Candemir”) Pages 1-10, 2021 and “Meta-learning with implicit gradients in a few-shot setting for medical image segmentation”, Khadka et al. (“Khadha”), pages 1-10, 1-12-2022 in further view of Karlinsky et al. (“Karlinsky” 20210256391 A1). Claim 11 : Fort, Candemir and Khadha disclose a method of claim 8, however may not explicitly disclose wherein the modified loss function comprises a negative log likelihood loss included an entropic regularization term to decrease the clustering distance from the cluster centroids. Karlinsky is provided because it discloses a cluster functionality and further utilizes negative log-likelihood for training and improving classification for clustering (Paragraph 40). Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate negative log-likelihood functionality for clustering in the modified Fort . One would have been motivated to provide the functionality because the method provides improved output (Karlinsky: Paragraph 40) . Claims 1 2 -1 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over “ Gaussian Prototypical Networks for Few-Shot Learning on Omniglot ”; Stanislav For t(“Fort”); pages 1-4, 2017 , “ Training Strategies for Radiology Deep Learning Models in Data-limited Scenarios ” Sema Candemir et al. (“Candemir”) Pages 1-10, 2021 and “Meta-learning with implicit gradients in a few-shot setting for medical image segmentation”, Khadka et al. (“Khadha”) , pages 1-10, 1-12-2022 in further view of “ Proceeding of International Conference on Computational Science and Applications ”, Shravani Nimbolkar et al. (“ Nimbolkar ”), pages 227-242, 2022 . Claim 1 2 : Fort, Candemir and Khadha disclose a method of claim 1, however may not disclose every feature including wherein each radiology classification task included in the meta-learning is associated with a corresponding domain that is different from the single domain associated with the first type of radiology report and the pre-training . Nimbolkar is provided because it discloses a meta-learning functionality (Page 228, Section 2.2) and incorporates different domain datasets (Page 229, Section 2.4-lungs/knee). Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate a variety of datasets as part of the meta-learning for the modified Fort . One would have been motivated to provide additional datasets because it expands the analysis and operability of a system , and can reduce overfitting . Claim 1 3 : Fort, Candemir and Khadha disclose a method of claim 1, wherein: the first radiology classification task and the first type of radiology report are associated with training the prototypical machine learning network using few-shot learning ( Fort: abstract ; few shot and Figure 1 classification task and Candemir: Figure 1; radiology training ) ; however may not disclose every feature including the second radiology classification task is different from the first radiology classification task associated with training the prototypical machine learning network; and the second type of radiology report is different from the first type of radiology report associated with training the prototypical machine learning network. Nimbolkar is provided because it discloses a meta-learning functionality (Page 228, Section 2.2) and incorporates different machine learning models (Page 230, Section 3.2 -Transfer learning and Siamese NN ) for different radiology data ( Nimbolkar: Page 229, Section 2.4 ) . Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate different radiology datasets as part of the meta-learning for the modified Fort . One would have been motivated to provide additional datasets because it expands the analysis and operability of a system, and can reduce overfitting. Claim 1 4 : Fort, Candemir and Khadha disclose a method of claim 1, but may not explicitly disclose wherein the radiology classification tasks included in the meta-learning includes one or more of : a lung nodule cancer screening classification task, having a first prototype representation corresponding to a high-risk classification and a second prototype representation corresponding to a not high-risk classification; a knee anterior cruciate ligament (ACL) classification task, having a first prototype representation corresponding to a tear classification and a second prototype representation corresponding to a normal classification; and a cervical spine classification task, having a first prototype representation corresponding to a normal classification and a second prototype representation corresponding to an abnormal classification. Nimbolkar is provided because it discloses a meta-learning functionality (Page 228, Section 2.2) and incorporates radiology datasets including lung datasets as part of a classification ( Page 225, Introduction and Page 229, Section 2.4 ) . Therefore it would have been obvious to one having ordinary skill in the art before the effective filling date of the claimed invention to apply a known technique to a known device ready for improvement and incorporate lung datasets as part of the meta-learning for Fort . One would have been motivated to provide additional datasets because it expands the analysis and classification operability of a system . Claim 1 5 : Fort, Candemir , Khadha and Nimbolkar method of claim 14, wherein the knee ACL classification tasks comprises one of: a knee ACL acute tear classification task, having a first prototype representation corresponding to an acute tear classification and a second prototype representation corresponding to a non-acute tear classification; or a knee ACL complete tear classification task, having a first prototype representation corresponding to a complete tear classification and a second prototype representation corresponding to a non- complete tear classification (Nimbolkar: Page 229, Section 2.4; Knee dataset) . Conclusion The prior art made of record and not relied upon is considered pertinent to Applicant’s disclosure: 20230113643 A1 0085-0089 “Prototypical Networks for Few-shot Learning” Snell et al. (“Snell”), pages 1-12, 2017. Soft Classification with Gaussian Mixture Model for Clinical Dual-Energy CT Reconstructions , Ruoqiao Zhang Meta-Learning for Medical Image Classification , Shi Hu Applicant is required under 37 C.F.R. § 1.111(c) to consider these references fully when responding to this action. It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck , 699 F.2d 1331, 1332-33, 216 U.S.P.Q. 1038, 1039 (Fed. Cir. 1983) (quoting In re Lemelson , 397 F.2d 1006, 1009, 158 U.S.P.Q. 275, 277 (C.C.P.A. 1968)). In the interests of compact prosecution, Applicant is invited to contact the examiner via electronic media pursuant to USPTO policy outlined MPEP § 502.03. All electronic communication must be authorized in writing. Applicant may wish to file an Internet Communications Authorization Form PTO/SB/439. Applicant may wish to request an interview using the Interview Practice website: http://www.uspto.gov/patent/laws-and-regulations/interview-practice. Applicant is reminded Internet e-mail may not be used for communication for matters under 35 U.S.C. § 132 or which otherwise require a signature. A reply to an Office action may NOT be communicated by Applicant to the USPTO via Internet e-mail. If such a reply is submitted by Applicant via Internet e-mail, a paper copy will be placed in the appropriate patent application file with an indication that the reply is NOT ENTERED. See MPEP § 502.03(II). Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHERROD KEATON whose telephone number is 571-270-1697. The examiner can normally be reached 9:30am to 5:00pm. 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 MICHELLE BECHTOLD can be reached at 571-431-0762. 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. /SHERROD L KEATON/ Primary Examiner, Art Unit 2148 3-10-2026