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 .
Notice to Applicants
This action is in response to the Response After Final Action filed on 01/16/2026.
Claims 1-2 and 4-21 are pending.
Corrective Actions by Applicant
Claims 12 and 20 have been amended.
Response to Arguments
The examiner has fully considered Applicant’s presented arguments.
As agreed upon in the interview on 12/09/2025, the use of the prior art reference Liao against the present Application would be limited to the teachings of U.S. Provisional Application 62/845,922, to which Liao claims priority. Respectively, said Provisional Application would fail to account for the deficiencies of Brestel regarding independent claims 1, 9, and 12. Thus, all previous 35 U.S.C. 103 rejections have been withdrawn.
However, after an updated search, new 103 rejections are presented below. As these rejections are not necessitated by claim amendments, this action is made non-final.
Claim Rejections – 35 USC § 103
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, 5-7, 9-10, 12-13, 15, and 20-21 are rejected under 35 U.S.C. 103 as being unpatentable over Brestel et al. (U.S. Publ. US-2019/0340753-A1) in view of Gleich et al. (U.S. Publ. US-2020/0250829-A1).
Regarding claim 1, Brestel discloses a method for detecting free intra-abdominal air (see figure 1), the method comprising:
receiving input data, including a first medical imaging data set of an abdominal region of a patient (see figure 1, step 104 and paragraphs 0092-0096, where a set of medical images are received, which can include abdominal images and 2D slices of 3D images from CT scans),
the first medical imaging data set including 3D imaging data (see paragraphs 0096-0097)
applying a trained function to the first medical imaging data set to generate output data (see figure 1, steps 106-110 and paragraphs 0105 and 0109-0111, where the images are input to a single-label neural network that outputs the likelihood of a visual finding being present in the input data, where the visual finding type can be pneumoperitoneum),
the trained function being trained based on training data including a plurality of training medical imaging data sets of respective abdominal regions of other patients (see figure 1, step 102, figure 4, and paragraphs 0085-0090, where the model is trained on anatomical datasets that can include abdominal images),
and each of image regions of the plurality of training medical imaging data sets containing free intra-abdominal air corresponding to an annotation (see paragraphs 0011-0012, where the training images are labeled according to the visual finding types, which can include pneumoperitoneum);
and providing the output data (see figure 1, step 110 and paragraph 0109, where the likelihood is output; see figure 1, step 116, where a patient can be diagnosed and/or treated using the visual finding output).
Brestel fails to disclose the first medical imaging data set including 3D imaging data subdivided into a plurality of image patches of a first size, and the plurality of image patches being adjacent to or overlapping one another (emphasis on the missing limitations added via underline).
Pertaining to the same field of endeavor, Gleich discloses the first medical imaging data set including 3D imaging data (see paragraph 0025) subdivided into a plurality of image patches (see figures 5-6 and paragraphs 0005, 0026, and 0091-0092, where each training image can be divided into groups of voxels using a center point of the image and orthogonal planes) of a first size (see figures 5-6 and paragraphs 0091-0093, where each of the groups of voxels are stated to be of a congruent size; in cases of non-spherical tumor images, the smallest sphere containing all tumor voxels can be used to ensure congruent sizes), and the plurality of image patches being adjacent to or overlapping one another (see figures 5-6 and paragraph 0081, where the groups of voxels are adjacent to one another, as they are evenly divided along the orthogonal planes).
Brestel and Gleich are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Gleich into Brestel because doing so allows for analysis of each individual patch/group for diagnosis (see Gleich paragraphs 0079-0080).
Regarding claim 5, Brestel in view of Gleich discloses wherein the trained function is based on a convolutional neural network or a deep neural network (see Brestel paragraph 0148).
Regarding claim 6, Brestel in view of Gleich discloses a system or apparatus configured to perform the method of claim 1 (see Brestel figure 2), the system or apparatus comprising:
a first interface configured to receive the input data (see Brestel figure 2, imaging interface 220 and paragraph 0071, where the computing device 204 can receive images using the imaging interface 220);
a computation unit configured to apply the trained function to generate the output data (see Brestel figure 2, memory 206 and trained single-label neural networks 222A; see Brestel paragraph 0073, where the memory 206 includes code 222A that executes the trained function);
and a second interface configured to provide the output data (see Brestel figure 2, computing device 204 and paragraph 0064, where the computing device can provide the data to client terminals 208).
Regarding claim 7, Brestel in view of Gleich discloses a non-transitory computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method of claim 1 (see Brestel paragraph 0052).
Regarding claim 9, Brestel discloses a method for providing a trained function for detecting free intra-abdominal air (see figure 1), the method comprising:
receiving first input training data including a first plurality of medical imaging data sets of an abdominal region, none among the first plurality of medical imaging data sets including an image regions containing free intra-abdominal air (see paragraph 0142, where a set of negative training data without abnormalities is obtained),
each among the first plurality of medical imaging data sets including first 3D imaging data (see paragraphs 0096-0097)
receiving second input training data including a second plurality of medical imaging data sets of the abdominal region, each among the second plurality of medical imaging data sets including at least one image region containing free intra-abdominal air (see paragraph 0141, where a set of positive training data with abnormalities is obtained),
each among the second plurality of medical imaging data sets including second 3D imaging data (see paragraphs 0096-0097)
training a function based on the first input training data and the second input training data to obtain a trained function (see paragraph 0144, where the network can be trained on a dataset made by combining the above negative and positive datasets);
and providing the trained function (see paragraph 0148).
Brestel fails to disclose each among the first plurality of medical imaging data sets including first 3D imaging data subdivided into a first plurality of image patches of a first size, and the first plurality of image patches being adjacent to or overlapping one another; and each among the second plurality of medical imaging data sets including second 3D imaging data subdivided into a second plurality of image patches of the first size, and the second plurality of image patches being adjacent to or overlapping one another (emphasis on the missing limitations added via underline).
Pertaining to the same field of endeavor, Gleich discloses each among the first plurality of medical imaging data sets including first 3D imaging data (see paragraph 0025) subdivided into a first plurality of image patches (see figures 5-6 and paragraphs 0005, 0026, and 0091-0092, where each training image can be divided into groups of voxels using a center point of the image and orthogonal planes) of a first size (see figures 5-6 and paragraphs 0091-0093, where each of the groups of voxels are stated to be of a congruent size; in cases of non-spherical tumor images, the smallest sphere containing all tumor voxels can be used to ensure congruent sizes), and the first plurality of image patches being adjacent to or overlapping one another (see figures 5-6 and paragraph 0081, where the groups of voxels are adjacent to one another, as they are evenly divided along the orthogonal planes);
each among the second plurality of medical imaging data sets including second 3D imaging data subdivided into a second plurality of image patches of the first size, and the second plurality of image patches being adjacent to or overlapping one another (see above citations from Gleich).
Brestel and Gleich are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Gleich into Brestel because doing so allows for analysis of each individual patch/group for diagnosis (see Gleich paragraphs 0079-0080).
Regarding claim 10, Brestel in view of Gleich discloses wherein the trained function is based on a convolutional neural network or a deep neural network (see Brestel paragraph 0148).
Regarding claim 12, Brestel discloses a training system for training a function for detecting free intra-abdominal air (see figure 2), the training system comprising:
a first training interface configured to receive input training data, the input training data including a plurality of medical imaging data sets of an abdominal region (see figure 2, imaging interface 220 and data interface 224; see paragraphs 0068-0071 and 0077-0078, where the computing device 204 can receive training images including abdominal images and associated labels using the imaging interface 220 and/or data interface 224),
each among the plurality of medical imaging data sets including 3D imaging data (see paragraphs 0096-0097)
a second training interface configured to receive output training data, the input training data being related to the output training data (see figure 2, imaging interface 220 and data interface 224; see paragraphs 0068-0071 and 0077-0078, where the computing device 204 can receive training images including abdominal images and associated labels using the imaging interface 220 and/or data interface 224),
the output training data including annotations indicating regions of free intra-abdominal air in the plurality of medical imaging data sets (see paragraphs 0011-0012, where the training images are labeled according to the visual finding types, which can include pneumoperitoneum);
a training computation unit configured to train a function based on the input training data and the output training data to obtain a trained function (see figure 2, memory 206 and training code 206B; see paragraph 0073, where the memory 206 includes training code 206B that executes the training steps of figure 4);
and a third training interface configured to provide the trained function (see figure 2, computing device 204 and client terminals 208; see paragraph 0064, where the computing device 204 can provide the trained function to client terminals 208).
Brestel fails to disclose each among the plurality of medical imaging data sets including 3D imaging data subdivided into a plurality of image patches of a first size, and the plurality of image patches being adjacent to or overlapping one another (emphasis on the missing limitations added via underline).
Pertaining to the same field of endeavor, Gleich discloses each among the plurality of medical imaging data sets including 3D imaging data (see paragraph 0025) subdivided into a plurality of image patches (see figures 5-6 and paragraphs 0005, 0026, and 0091-0092, where each training image can be divided into groups of voxels using a center point of the image and orthogonal planes) of a first size (see figures 5-6 and paragraphs 0091-0093, where each of the groups of voxels are stated to be of a congruent size; in cases of non-spherical tumor images, the smallest sphere containing all tumor voxels can be used to ensure congruent sizes), and the plurality of image patches being adjacent to or overlapping one another (see figures 5-6 and paragraph 0081, where the groups of voxels are adjacent to one another, as they are evenly divided along the orthogonal planes).
Brestel and Gleich are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Gleich into Brestel because doing so allows for analysis of each individual patch/group for diagnosis (see Gleich paragraphs 0079-0080).
Regarding claim 13, Brestel in view of Gleich discloses a non-transitory computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method of claim 10 (see Brestel paragraph 0148).
Regarding claim 15, Brestel in view of Gleich discloses a system or apparatus configured to perform the method of claim 5 (see Brestel figure 2), the system or apparatus comprising:
a first interface configured to receive the input data (see Brestel figure 2, imaging interface 220 and paragraph 0071, where the computing device 204 can receive images using the imaging interface 220);
a computation unit configured to apply the trained function to generate the output data (see Brestel figure 2, memory 206 and trained single-label neural networks 222A; see Brestel paragraph 0073, where the memory 206 includes code 222A that executes the trained function);
and a second interface configured to provide the output data (see Brestel figure 2, computing device 204 and paragraph 0064, where the computing device can provide the data to client terminals 208).
Regarding claim 20, Brestel in view of Gleich discloses scanning the patient using a medical imaging scanner to obtain a raw data set (see Brestel figure 1, step 104 and paragraphs 0093-0097, where a set of medical images are captured and received, which can include abdominal images and 2D slices of 3D images from CT scans);
and reconstructing the raw data set to obtain the first medical imaging data set (see Brestel figure 1, steps 106-108 and paragraphs 0098-0108, where a variety of processing is performed to obtain an instance of a reconstructed dataset, such as neural network processing, filtering, and classification),
wherein the output data includes an indication indicating a presence or an absence of free intra-abdominal air (see Brestel figure 1, step 110 and paragraphs 0109-0111, where the images are input to a single-label neural network that outputs the likelihood of a visual finding being present in the input data, where the visual finding type can be pneumoperitoneum),
and the providing the output data includes displaying the first medical imaging data set with the indication (see Brestel figure 1, steps 112-114 and paragraphs 0112-0117, where a priority/triage list indicating the identified datasets is output to a user).
Regarding claim 21, Brestel fails to disclose the limitations of claim 21.
Pertaining to the same field of endeavor, Gleich discloses subdividing the 3D imaging data into the plurality of image patches of first pixel dimensions, the first pixel dimensions corresponding to the first size (see figures 5-6 and paragraphs 0091-0093, where each of the groups of voxels are stated to be of a congruent size; in cases of non-spherical tumor images, the smallest sphere containing all tumor voxels can be used to ensure congruent sizes).
Brestel and Gleich are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Gleich into Brestel because doing so allows for analysis of each individual patch/group for diagnosis (see Gleich paragraphs 0079-0080).
Claims 2, 4, 8, 11, 14, and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Brestel et al. (U.S. Publ. US-2019/0340753-A1) in view of Gleich et al. (U.S. Publ. US-2020/0250829-A1), and further in view of Nye et al. (U.S. Publ. US-2020/0211694-A1).
Regarding claim 2, Brestel in view of Gleich fails to teach the limitations of claim 2.
Pertaining to the same field of endeavor, Nye discloses wherein the annotation includes a marker that indicates a position inside a region of free intra-abdominal air (see paragraph 0107, where training images depicting a pneumothorax in a lung image, for example, are annotated with location information of the pneumothorax; note that paragraphs 0145-0146 specify that detecting pneumoperitoneum is also in the scope of the invention).
Brestel and Nye are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Nye into Brestel and Gleich because labeling the features of training data allows for the model to better learn the features (see Nye paragraphs 0034, 0051, and 0053).
Regarding claim 4, Brestel in view of Gleich fails to teach the limitations of claim 4.
Pertaining to the same field of endeavor, Nye discloses wherein the applying the trained function includes determining whether one among the plurality of image patches contains free intra-abdominal air (see figure 8, steps 840-860 and paragraph 0085, where each patch is fed into the trained network for feature processing).
Brestel and Nye are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Nye into Brestel and Gleich because one of ordinary skill in the art would recognize that Nye's method of dividing input images into patches is a common method of enabling a model to focus more closely on individual image features for feature learning.
Regarding claim 8, Brestel in view of Gleich fails to teach the limitations of claim 8.
Pertaining to the same field of endeavor, Nye discloses a non-transitory computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method of claim 2 (see paragraph 0106).
Brestel and Nye are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Nye into Brestel and Gleich because labeling the features of training data allows for the model to better learn the features (see Nye paragraphs 0034, 0051, and 0053).
Regarding claim 11, Brestel in view of Gleich fails to teach the limitations of claim 11.
Pertaining to the same field of endeavor, Nye discloses wherein each among the second plurality of medical imaging data sets includes annotation information indicating the at least one image region containing free intra-abdominal air (see paragraph 0107, where training images depicting a pneumothorax in a lung image, for example, are annotated with location information of the pneumothorax; note that paragraphs 0145-0146 specify that detecting pneumoperitoneum is also in the scope of the invention).
Brestel and Nye are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Nye into Brestel and Gleich because labeling the features of training data allows for the model to better learn the features (see Nye paragraphs 0034, 0051, and 0053).
Regarding claim 14, Brestel in view of Gleich fails to teach the limitations of claim 14.
Pertaining to the same field of endeavor, Nye discloses a non-transitory computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method of claim 11 (see paragraph 0106).
Brestel and Nye are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Nye into Brestel and Gleich because labeling the features of training data allows for the model to better learn the features (see Nye paragraphs 0034, 0051, and 0053).
Regarding claim 16, Brestel in view of Gleich and Nye discloses claim 16 as applied to claim 11 above.
Claims 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Brestel et al. (U.S. Publ. US-2019/0340753-A1) in view of Gleich et al. (U.S. Publ. US-2020/0250829-A1), and further in view of Schieke (U.S. Publ. US-2018/0005417-A1).
Regarding claim 17, Brestel in view of Gleich fails to disclose the limitations of claim 17. More specifically, Liao only discloses dividing the dataset into a single set of patches using a single patch size (see paragraph 0038, where the square window used to divide the image into patches has a fixed 50 x 50 pixel size corresponding to a first size).
Pertaining to the same field of endeavor, Schieke discloses wherein the plurality of image patches is a first plurality of image patches; and the first medical imaging data set includes the 3D imaging data subdivided into the first plurality of image patches and a second plurality of image patches, the second plurality of image patches being of a second size (first see paragraphs 0065 and 0076-0078, where three-dimensional medical image datasets are obtained for super-resolution and disease detection; the datasets can include individual patents before and after treatment; then see paragraphs 0094-0096, where moving windows are applied to each image of each dataset to divide the images into a discrete grid of voxels/patches to individually analyze; as multiple moving windows of different sizes can be defined, multiple pluralities of patches of different sizes are thus obtained).
Brestel and Schieke are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Schieke into Brestel and Gleich because doing so allows for varying the data collected from each moving window (see Schieke paragraph 0095).
Regarding claim 18, Brestel in view of Gleich fails to disclose the limitations of claim 18. More specifically, Liao only discloses dividing the dataset into a single set of patches using a single patch size (see paragraph 0038, where the square window used to divide the image into patches has a fixed 50 x 50 pixel size corresponding to a first size).
Pertaining to the same field of endeavor, Schieke discloses wherein the first plurality of medical imaging data sets include the first 3D imaging data subdivided into the first plurality of image patches and a third plurality of image patches, the third plurality of image patches being of a second size; and the second plurality of medical imaging data sets include the second 3D imaging data subdivided into the second plurality of image patches and a fourth plurality of image patches, the fourth plurality of image patches being of the second size (first see paragraphs 0065 and 0076-0078, where three-dimensional medical image datasets are obtained for super-resolution and disease detection; the datasets can include individual patents before and after treatment; then see paragraphs 0094-0096, where moving windows are applied to each image of each dataset to divide the images into a discrete grid of voxels/patches to individually analyze; as multiple moving windows of different sizes can be defined, multiple pluralities of patches of different sizes are thus obtained).
Brestel and Schieke are considered analogous art, as they are both directed to neural networks for detecting diseases in radiology images. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Schieke into Brestel and Gleich because doing so allows for varying the data collected from each moving window (see Schieke paragraph 0095).
Regarding claim 19, Brestel in view of Gleich and Schieke discloses claim 19 as applied to claim 17 above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICHOLAS JOHN HELCO whose telephone number is (703)756-5539. The examiner can normally be reached on Monday-Friday from 9:00 AM to 5:00 PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Matthew Bella, can be reached at telephone number 571-272-7778. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from Patent Center. Status information for published applications may be obtained from Patent Center. Status information for unpublished applications is available through Patent Center for authorized users only. Should you have questions about access to Patent Center, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
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) Form at https://www.uspto.gov/patents/uspto-automated- interview-request-air-form.
/NICHOLAS JOHN HELCO/Examiner, Art Unit 2667
/MATTHEW C BELLA/Supervisory Patent Examiner, Art Unit 2667