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 .
Response to Arguments
Applicant’s arguments, see Applicants Remarks pages 10-11, filed 04/01/26, with respect to the rejection(s) of claim(s) 2, 3, 5-12, 14, 15, 17-19 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Wang et al US 20190362494.
Regarding claim 2, Examiner indicated that claim 4 would be allowable if written in independent form. Upon further search and consideration, the claim is now rejected. Wang et al teaches of training a deep learning model, using acquired sequences of image patches to predict the sequence of blood vessel (fig 1a and 1b). This would read on providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network and training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 02/10/26 was filed. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim 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 (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
Claim(s) 2, 3, 5, 6, 8-12, 14, 15, 17, 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Anand et al US 20230111306 in view of Bovik et al US 20180247127 further in view of Yang et al US 20200302176 further in view of Wang et al US 20190362494.
Regarding claim 2, Anand et al teaches a self-supervised machine learning method for learning visual representations in medical images (abstract), comprising:
receiving a plurality of medical images of similar anatomy (the medical images in the set can include images depicting same or different anatomical regions and patients, captured using same or different imaging modalities (e.g., XR, PT, MRI, CT, etc.), and/or captured using same or different acquisition parameters/protocols (paragraph 0052);
dividing each of the plurality of medical images into its own sequence of non-overlapping patches, wherein a unique portion of each medical image appears in each patch in the sequence of non-overlapping patches (an input image 202 and image is cut into several different patches 204 (dividing images into patches). These patches are linear projections that contain different portions of the input image 202. The patches and the relative positions (sequences) of the patches are fed into the transformer encoder 208 which learns patch feature representations 210 for each of the patches as well as an intermediate a representation for the entire input image. To facilitate this end, the ViT employs a plurality of different attention heads 206 that respectively identify the regions (unique portion) in the respective patches that contain the most important/influential features to be fed into the transformer encoder 208 (paragraph 0058);
learning, via a vision transformer network, patch-wise high-level contextual features in the plurality of medical images (vision transformer (ViT) network 200 that can be used as the transformer network 116 to generate feature representations for medical images. To facilitate this end, the ViT employs a plurality of different attention heads 206 that respectively identify the regions (unique portion) in the respective patches that contain the most important/influential features to be fed into the transformer encoder 208 (paragraph 0058); and
Anand et al fails to teach randomizing the sequence of non-overlapping patches for each of the plurality of medical images;
randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images;
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network
Bovik et al teaches randomizing the sequence of non-overlapping patches for each of the plurality of medical images (Each patch was classified using multiple probabilistic SVCs, one per distortion type, to determine the likelihood that the patch belonged to that distorted class or to the natural image class (paragraph 0145) Note: patches are randomly sequence based on the distorted class;
randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images (Each patch was classified using multiple probabilistic SVCs, one per distortion type, to determine the likelihood that the patch belonged to that distorted class or to the natural image class (paragraph 0145);
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network (given these two distorted sets, the dataset was split into two non-overlapping subsets: 80 percent for training and 20 percent for testing. A Support Vector Classifier (SVC) (vision transformer network) was used to map the features between two classes. Random 80/20 splits were produced and classified (paragraph 0127) Each patch was classified using multiple probabilistic SVCs (paragraph 0145); and
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al to include: randomizing the sequence of non-overlapping patches for each of the plurality of medical images; randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Anand et al in view of Bovik et al fails to teach learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images
Yang et al teaches learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images (the generative portion generates image data during training. The generated image data comprises various generated representations of the feature being re-identified, in which each variation changes the appearance of the feature in some way. This approach may permit the neural network to learn to recognize fine-grained aspects of the features being analyzed (paragraph 0015).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al to include: learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Anand et al in view of Bovik et al further in view of Yang et al fails to teach training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images
Wang et al teaches training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images (training a deep learning model, using acquired sequences of image patches to predict the sequence of blood vessel (fig 1a and 1b)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Regarding claim 3, Anand et al in view of Bovik et al further in view of Yang et al teaches wherein learning, via a vision transformer network, patch- wise high-level contextual features comprises learning high-level anatomical structures and their relative relationships in the plurality of medical images (Anand et al: Image set 601 corresponds to one image while image set 602 corresponds to the second image. The upper row of images in each set corresponds to the attention head map while the lower row of images depicts the attention head maps overlaid onto the original input image. Both images were processed by the ViT network 200 to generate the corresponding attention head maps which are overlaid onto the original input images. As illustrated in FIG. 6, for both input images, the ViT network 200 attention head maps consistently point to similar anatomical regions of interest across the different images, which in this case include the bone region of the knee (paragraph 0077 and fig 6).
Regarding claim 5, Anand et al in view of Bovik et al further in view of Yang et al further in view of Wang et al teaches wherein training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images comprises training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images based on an appearance of each patch in the sequence of non-overlapping patches for each of the plurality of medical images (Wang et al: the corresponding sequence(s) of image patches and the sequence(s) of blood vessel condition parameters may be obtained as training data. The corresponding sequence(s) of blood vessel condition parameters may be that predicted previously for the patient's blood vessel path by using the deep learning model (paragraph 0029).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: wherein training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images comprises training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images based on an appearance of each patch in the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Regarding claim 6, Anand et al in view of Bovik et al further in view Yang et al further in view of Wang et al teaches wherein learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images comprises learning details in texture variations embedded throughout an entirety of the plurality of medical images (Yang et al: the generative portion generates image data during training. The generated image data comprises various generated representations of the feature being re-identified, in which each variation changes the appearance of the feature in some way. This approach may permit the neural network to learn to recognize fine-grained aspects of the features being analyzed (paragraph 0015)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al to include: wherein learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images comprises learning details in texture variations embedded throughout an entirety of the plurality of medical images.
The reason for doing so would be to learning different changes in an image in order to accurately analyze medical images.
Regarding claim 8, Anand et al teaches a system comprising:
a memory to store instructions (memory (paragraph 0010); and
a processor to execute the instructions stored in the memory (processor (paragraph 0010);
wherein the system is specially configured to execute instructions via the processor for performing the following operations:
receiving a plurality of medical images of similar anatomy (the medical images in the set can include images depicting same or different anatomical regions and patients, captured using same or different imaging modalities (e.g., XR, PT, MRI, CT, etc.), and/or captured using same or different acquisition parameters/protocols (paragraph 0052);
dividing each of the plurality of medical images into its own sequence of non- overlapping patches, wherein a unique portion of each medical image appears in each patch in the sequence of non-overlapping patches (an input image 202 and image is cut into several different patches 204 (dividing images into patches). These patches are linear projections that contain different portions of the input image 202. The patches and the relative positions (sequences) of the patches are fed into the transformer encoder 208 which learns patch feature representations 210 for each of the patches as well as an intermediate a representation for the entire input image. To facilitate this end, the ViT employs a plurality of different attention heads 206 that respectively identify the regions (unique portion) in the respective patches that contain the most important/influential features to be fed into the transformer encoder 208 (paragraph 0058);
learning, via a vision transformer network, patch-wise high-level contextual features in the plurality of medical images (vision transformer (ViT) network 200 that can be used as the transformer network 116 to generate feature representations for medical images. To facilitate this end, the ViT employs a plurality of different attention heads 206 that respectively identify the regions (unique portion) in the respective patches that contain the most important/influential features to be fed into the transformer encoder 208 (paragraph 0058); and
Anand et al fails teach randomizing the sequence of non-overlapping patches for each of the plurality of medical images;
randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images;
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network
Bovik et al teaches randomizing the sequence of non-overlapping patches for each of the plurality of medical images (Each patch was classified using multiple probabilistic SVCs, one per distortion type, to determine the likelihood that the patch belonged to that distorted class or to the natural image class (paragraph 0145) Note: patches are randomly sequence based on the distorted class;
randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images (Each patch was classified using multiple probabilistic SVCs, one per distortion type, to determine the likelihood that the patch belonged to that distorted class or to the natural image class (paragraph 0145);
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network (given these two distorted sets, the dataset was split into two non-overlapping subsets: 80 percent for training and 20 percent for testing. A Support Vector Classifier (SVC) (vision transformer network) was used to map the features between two classes. Random 80/20 splits were produced and classified (paragraph 0127) Each patch was classified using multiple probabilistic SVCs (paragraph 0145)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al to include: randomizing the sequence of non-overlapping patches for each of the plurality of medical images; randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images; providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network
The reason for doing so would be to accurately analyze medical images.
Anand et al in view of Bovik et al fails to teach learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images
Yang et al teaches learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images (the generative portion generates image data during training. The generated image data comprises various generated representations of the feature being re-identified, in which each variation changes the appearance of the feature in some way. This approach may permit the neural network to learn to recognize fine-grained aspects of the features being analyzed (paragraph 0015).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al to include: learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Anand et al in view of Bovik et al further in view of Yang et al fails to teach training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images
Wang et al teaches training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images (training a deep learning model, using acquired sequences of image patches to predict the sequence of blood vessel (fig 1a and 1b)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Regarding claim 9, Anand et al in view of Bovik et al further in view of Yang et al teaches wherein learning, via a vision transformer network, patch- wise high-level contextual features comprises learning high-level anatomical structures and their relative relationships in the plurality of medical images (Anand et al: Image set 601 corresponds to one image while image set 602 corresponds to the second image. The upper row of images in each set corresponds to the attention head map while the lower row of images depicts the attention head maps overlaid onto the original input image. Both images were processed by the ViT network 200 to generate the corresponding attention head maps which are overlaid onto the original input images. As illustrated in FIG. 6, for both input images, the ViT network 200 attention head maps consistently point to similar anatomical regions of interest across the different images, which in this case include the bone region of the knee (paragraph 0077 and fig 6).
Regarding claim 10, Anand et al teaches wherein learning, via the vision transformer network, patch-wise high-level contextual features in the plurality of medical images comprises:
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network (Bovik et al: given these two distorted sets, the dataset was split into two non-overlapping subsets: 80 percent for training and 20 percent for testing. A Support Vector Classifier (SVC) (vision transformer network) was used to map the features between two classes. Random 80/20 splits were produced and classified (paragraph 0127) Each patch was classified using multiple probabilistic SVCs (paragraph 0145); and
training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images (Wang et al: training a deep learning model, using acquired sequences of image patches to predict the sequence of blood vessel (fig 1a and 1b).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network; training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images
The reason for doing so would be to accurately analyze medical images.
Regarding claim 11, Anand et al in view of Bovik et al further in view of Yang et al further in view of Wang et al teaches wherein training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images comprises training the vision transformer network to predict the sequence of non- overlapping patches for each of the plurality of medical images based on an appearance of each patch in the sequence of non-overlapping patches for each of the plurality of medical images (Wang et al: the corresponding sequence(s) of image patches and the sequence(s) of blood vessel condition parameters may be obtained as training data. The corresponding sequence(s) of blood vessel condition parameters may be that predicted previously for the patient's blood vessel path by using the deep learning model (paragraph 0029).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: wherein training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images comprises training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images based on an appearance of each patch in the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Regarding claim 12, Anand et al in view of Bovik et al further in view Yang et al teaches wherein learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images comprises learning details in texture variations embedded throughout an entirety of the plurality of medical images (Yang et al: the generative portion generates image data during training. The generated image data comprises various generated representations of the feature being re-identified, in which each variation changes the appearance of the feature in some way. This approach may permit the neural network to learn to recognize fine-grained aspects of the features being analyzed (paragraph 0015)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al to include: wherein learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images comprises learning details in texture variations embedded throughout an entirety of the plurality of medical images.
The reason for doing so would be to learning different changes in an image in order to accurately analyze medical images.
Regarding claim 14, Anand et al teaches a non-transitory computer readable storage media having instructions stored thereupon that (paragraph 0048), when executed by a process of a system specially configured for diagnosing disease within new medical images (a disease classification for the medial image, and so on. The computer executable components can further comprise a classification component that determines a classification of the new medical image based on the one or more defined class attributes associated with the one or more similar medical images (paragraph 0014);
wherein the instructions cause the system to perform operations including:
receiving a plurality of medical images of similar anatomy (the medical images in the set can include images depicting same or different anatomical regions and patients, captured using same or different imaging modalities (e.g., XR, PT, MRI, CT, etc.), and/or captured using same or different acquisition parameters/protocols (paragraph 0052);
dividing each of the plurality of medical images into its own sequence of non-overlapping patches, wherein a unique portion of each medical image appears in each patch in the sequence of non-overlapping patches (an input image 202 and image is cut into several different patches 204 (dividing images into patches). These patches are linear projections that contain different portions of the input image 202. The patches and the relative positions (sequences) of the patches are fed into the transformer encoder 208 which learns patch feature representations 210 for each of the patches as well as an intermediate a representation for the entire input image. To facilitate this end, the ViT employs a plurality of different attention heads 206 that respectively identify the regions (unique portion) in the respective patches that contain the most important/influential features to be fed into the transformer encoder 208 (paragraph 0058);
learning, via a vision transformer network, patch-wise high-level contextual features in the plurality of medical images (vision transformer (ViT) network 200 that can be used as the transformer network 116 to generate feature representations for medical images. To facilitate this end, the ViT employs a plurality of different attention heads 206 that respectively identify the regions (unique portion) in the respective patches that contain the most important/influential features to be fed into the transformer encoder 208 (paragraph 0058); and
Anand et al fails to teach randomizing the sequence of non-overlapping patches for each of the plurality of medical images;
randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images;
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network
Bovik et al teaches randomizing the sequence of non-overlapping patches for each of the plurality of medical images (Each patch was classified using multiple probabilistic SVCs, one per distortion type, to determine the likelihood that the patch belonged to that distorted class or to the natural image class (paragraph 0145) Note: patches are randomly sequence based on the distorted class;
randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images (Each patch was classified using multiple probabilistic SVCs, one per distortion type, to determine the likelihood that the patch belonged to that distorted class or to the natural image class (paragraph 0145);
providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network (given these two distorted sets, the dataset was split into two non-overlapping subsets: 80 percent for training and 20 percent for testing. A Support Vector Classifier (SVC) (vision transformer network) was used to map the features between two classes. Random 80/20 splits were produced and classified (paragraph 0127). Each patch was classified using multiple probabilistic SVCs (paragraph 0145)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al to include: randomizing the sequence of non-overlapping patches for each of the plurality of medical images; randomly distorting the unique portion of each medical image that appears in each patch in the sequence of non-overlapping patches for each of the plurality of medical images; providing the randomized sequence of non-overlapping patches for each of the plurality of medical images to the vision transformer network
The reason for doing so would be to accurately analyze medical images.
Anand et al in view of Bovik et al fails to teach learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images
Yang et al teaches learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images (the generative portion generates image data during training. The generated image data comprises various generated representations of the feature being re-identified, in which each variation changes the appearance of the feature in some way. This approach may permit the neural network to learn to recognize fine-grained aspects of the features being analyzed (paragraph 0015).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al to include: learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Anand et al in view of Bovik et al further in view of Yang et al fails to teach training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images
Wang et al teaches training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images (training a deep learning model, using acquired sequences of image patches to predict the sequence of blood vessel (fig 1a and 1b)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Regarding claim 15, Anand et al in view of Bovik et al further in view Yang et al teaches wherein learning, via a vision transformer network, patch-wise high-level contextual features comprises learning high-level anatomical structures and their relative relationships in the plurality of medical images (Anand et al: Image set 601 corresponds to one image while image set 602 corresponds to the second image. The upper row of images in each set corresponds to the attention head map while the lower row of images depicts the attention head maps overlaid onto the original input image. Both images were processed by the ViT network 200 to generate the corresponding attention head maps which are overlaid onto the original input images. As illustrated in FIG. 6, for both input images, the ViT network 200 attention head maps consistently point to similar anatomical regions of interest across the different images, which in this case include the bone region of the knee (paragraph 0077 and fig 6).
Regarding claim 17, Anand et al in view of Bovik et al further in view of Yang et al further in view of Wang et al teaches wherein training the vision transformer network to predict the sequence of non- overlapping patches for each of the plurality of medical images comprises training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images based on an appearance of each patch in the sequence of non-overlapping patches for each of the plurality of medical images (Wang et al: the corresponding sequence(s) of image patches and the sequence(s) of blood vessel condition parameters may be obtained as training data. The corresponding sequence(s) of blood vessel condition parameters may be that predicted previously for the patient's blood vessel path by using the deep learning model (paragraph 0029).
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al further in view of Yang et al to include: wherein training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images comprises training the vision transformer network to predict the sequence of non-overlapping patches for each of the plurality of medical images based on an appearance of each patch in the sequence of non-overlapping patches for each of the plurality of medical images.
The reason for doing so would be to accurately analyze medical images.
Regarding claim 18, Anand et al in view of Bovik et al further in view Yang et al teaches wherein learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images comprises learning details in texture variations embedded throughout an entirety of the plurality of medical images (Yang et al: the generative portion generates image data during training. The generated image data comprises various generated representations of the feature being re-identified, in which each variation changes the appearance of the feature in some way. This approach may permit the neural network to learn to recognize fine-grained aspects of the features being analyzed (paragraph 0015)
Therefore, it would have been obvious to a person of ordinary skill in the art to modify Anand et al in view of Bovik et al to include: wherein learning simultaneously, via the vision transformer network, fine-grained features embedded in the plurality of medical images comprises learning details in texture variations embedded throughout an entirety of the plurality of medical images.
The reason for doing so would be to learning different changes in an image in order to accurately analyze medical images.
Allowable Subject Matter
Claims 7 and 19 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL L BURLESON whose telephone number is (571)272-7460.
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, Akwasi Sarpong can be reached on 571 270-3438 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.
Michael Burleson
Patent Examiner
Art Unit 2683
Michael Burleson
June 26, 2026
/MICHAEL BURLESON/
/AKWASI M SARPONG/SPE, Art Unit 2681 6/29/2026
Prosecution Insights