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 .
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.
Claim(s) 1-7 is/are rejected under 35 U.S.C. 103 as being unpatentable over () in view of Kim (KR 102100699 B1) in view of Achmad (CN 111247595 A), and further in view of Lu (CN 110838108 A).
Regarding claim 1. Kim teaches a method for classifying for Alzheimer's disease with a neural network (16), the method comprising:
acquiring (40) at least a first type of image data representing a patient (see page 6, lines 20-23, the learning data set 200 is an image of a lesion region obtained by extracting a region in which a lesion appears in the diagnostic images 201, 202, and 203 and the diagnostic images 201, 202, and 203 of the patient having different diseases);
inputting (42) the first type of image data into the neural network (16), the neural network (16) having first and second input branches (30, 31) each including multiple layers, the first input branch (30) being for the first type of image data and the second input branch (31) being for a second type of image data (see page 7, lines 4-7, the lesion area integrated learning model 300 may be trained to detect the images 211, 212, and 213 of the lesion area corresponding to the diagnostic images 201, 202, and 203, and furthermore, the lesion area integrated learning model The 300 may be learned based on a convolutional neural network (CNN) technique or a pooling technique);
outputting (44) a classification of the patient with respect to Alzheimer's disease from the neural network (16) in response to the inputting (42), the neural network (16) having an output portion having been trained to output the classification in response to the inputting (42) of the first type of image data, the second type of image data, and both the first and the second type of image data (see page 9 and 10, lines 35 and 1-22, integrated learning model 610 may include a first integrated learning model 611, a second integrated learning model 612, and a third integrated learning model 613. At this time, the input of the first integrated learning model 611 is set to a first diagnostic image (eg, MRI of the prostate region of a patient with prostate cancer), and the target variable is a first lesion region image (eg, an image of the prostate cancer region). Can be set to Since the first integrated learning model 612 performs GAN-based learning, it is possible to generate real data and fake data corresponding to the first lesion region image. And the input of the second integrated learning model 612 is the second diagnostic image (eg, MRI of the brain region of a brain tumor patient) and the output value of the first integrated learning model 611 (actually corresponding to the first lesion region image) It can be set as (real) data and fake data, and the target variable can be set as a second lesion area image (eg, an image of a brain tumor area). Accordingly, the second integrated learning model 612 can construct a learning model that reflects both the first diagnostic image and the second diagnostic image, and the data considering both the first diagnostic image and the second diagnostic image, that is, the actual ( Real data and fake data can be output. In addition, the input of the third integrated learning model 613 includes a third diagnostic image (eg, MRI of the brain region of an Alzheimer's patient), and an output value of the second integrated learning model 612 (real data and fake ( fake) data), and the target variable may be set as a third lesion region image (eg, an image of the Alzheimer's expression region). After all, the third integrated learning model 613 can construct a learning model that reflects all of the first to third diagnostic images, and is configured to output data that detects a lesion area in consideration of both of the first to third diagnostic images; see page 12, lines 30-32, The disease severity detection unit 97 may include a severity integrated learning model 970, in which the severity integrated learning model 970 detects the type of disease and the severity of the disease in response to input of a uniform sized lesion region image).
However, Kim does not expressly teaches displaying (46) the classification on a display screen.
Achmad teaches to provide the first and second indication which subject with MCI and which subject has AD identification; a display, said display for displaying classification. computer system can be arranged to subject to classification is MCI provides MCI is further classification of the MCI early or advanced stage MCI (see page 11, lines 11-15).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Kim by Achmad for providing to provide the first and second indication which subject with MCI and which subject has AD identification; a display, said display for displaying classification, as displaying (46) the classification on a display screen. Therefore, combining the elements from prior arts according to known methods and technique, such as displaying classification, would yield predictable results.
However, the combination does not expressly teach image data representing at least a first volume of a patient.
Lu teaches that prediction model based on medical image prediction model building scheme based on medical image constructed solution provided by the embodiments of the present application compared to the characteristic current extracting artificial design, training the classifier to predict and model to forecast service 3D-CNN, Alzheimer ' s disease, frontotemporal dementia examination or prediction in the scene is more practical (see page 9, lines 29-33).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination by Lu for providing medical image constructed solution provided by the embodiments of the present application compared to the characteristic current extracting artificial design, training the classifier to predict and model to forecast service 3D-CNN, Alzheimer ' s disease, as image data representing at least a first volume of a patient. Therefore, combining the elements from prior arts according to known methods and technique would yield predictable results.
Regarding claim 2. The combination teaches the method of claim 1
wherein acquiring (40) comprises acquiring (40) the first type of image data and the second type of image data (see Kim, page 8 and 9, lines 35 and 1-3, the lesion region learning unit 13 inputs the first diagnostic image (eg, MRI of the prostate region of a patient with prostate cancer) and sets the first lesion region image (eg, the image of the prostate cancer region) as a target variable. 1 The learning of the integrated learning model 510 can be performed; see page 9, lines 6-8, the lesion area learning unit 13 uses a second diagnostic image (eg, MRI of a brain tumor patient's brain area) as an input to the second integrated learning model 520, and a second lesion area image (eg, a brain tumor area)),
wherein inputting (42) comprises inputting (42) the first type of image data into the first input branch (30) and inputting (42) the second type of image data into the second input branch (31) (see Kim, page 6, lines 18-25, the lesion integrated learning apparatus 10 may prepare a learning data set 200 (see FIG. 2) for performing learning of the above-described lesion area integrated learning model or severity integrated learning model. The learning data set 200 is an image of a lesion region obtained by extracting a region in which a lesion appears in the diagnostic images 201, 202, and 203 and the diagnostic images 201, 202, and 203 of the patient having different diseases. (211, 212, 213), and the type of disease corresponding to the lesion area (221a, 222a, 223a) and the severity of the disease (221b, 222b, 223b) indicating data (221, 222, 223) may be included have), and
wherein outputting (44) comprises outputting (44) the classification in response to the inputting (42) of both the first and second types of image data (see Kim, page 8, lines 6-15, the lesion area learning unit 13 uses a second diagnostic image (eg, MRI of a brain tumor patient's brain area) as an input to the second integrated learning model 520, and a second lesion area image (eg, a brain tumor area). Image) can be set as an objective variable, and at this time, additionally, the output value of the first integrated learning model 510, that is, real data and fake data corresponding to the first lesion area image. Can be set together as an input of the second integrated learning model 520. Accordingly, the second integrated learning model 520 can construct a learning model that reflects both the first diagnostic image and the second diagnostic image, and the data considering both the first diagnostic image and the second diagnostic image, that is, the actual ( Real data and fake data can be output).
Regarding claim 3. The combination teaches the method of claim 2
wherein acquiring (40) comprises acquiring (40) the first type of image data as a first type of positron emission tomography data and acquiring (40) the second type of image data as a second type of positron emission tomography data or magnetic resonance data (see Lu page, lines , Medical imaging device can be common in medicine by the image collection device, for example, a magnetic resonance imaging device, a positron emission computed tomography device, computer tomography scanning, CT (Computed Tomography) device).
Regarding claim 4. The combination teaches the method of claim 1
wherein acquiring (40) comprises acquiring (40) the first type of image data as volume data representing a three-dimensional region of the patient, and wherein inputting (42) comprises inputting (42) the volume data to the first input branch (30) (see page 9, lines 29-33, prediction model based on medical image prediction model building scheme based on medical image constructed solution provided by the embodiments of the present application compared to the characteristic current extracting artificial design, training the classifier to predict and model to forecast service 3D-CNN, Alzheimer ' s disease, frontotemporal dementia examination or prediction in the scene is more practical).
Regarding claim 5. The combination teaches the method of claim 1 wherein inputting (42) comprises inputting (42) where the multiple layers of each of the first and second input branches (30, 31) comprise convolutional neural network layers (24) (see Kim, page 15, lines 8-12, in the lesion area integrated learning model 300 or the severity integrated learning model 400 of the present disclosure, a convolutional neural network may be used to extract "features" such as borders, line colors, and the like from input data (images). It may include a plurality of layers (layers). Each layer may receive input data and process input data of the corresponding layer to generate output data).
Regarding claim 6. The combination teaches the method of claim 1 wherein outputting (44) comprises outputting (44) the classification as a prediction for conversion of the patient from mild cognitive impairment to Alzheimer's disease (see Lu, page 8, lines 15-22, database and not strictly limited only can provide the first image to be predicted, which can also provide a second image to be predicted. research shows that although Alzheimer ' s disease can cause atrophy of the whole brain, but hippocampus (hippcampus) is the area of damaged at first, in one embodiment, when the Alzheimer ' s disease prediction, the electronic device obtains the patient brain region image collected by the medical imaging device, it can with higher precision the hippocampus area divided, and outputting the classification prediction result by the prediction model based on medical image pre-constructed).
Regarding claim 7. The combination teaches the method of claim 1 further comprising inputting (42) cognitive test information for the patient to the neural network (16), and wherein outputting (44) comprises outputting (44) the classification in response to the inputting (42) of the first type of image data and the inputting (42) of the cognitive test information (see Lu, page 13, lines 5-9, the first image input to be predicted based on multi-target learning depth neural network feature extraction sub-network to down-sampling operation, extracting the first characteristic information of a prediction image, and the depth of the classifier neural network extracting feature information of the input based on the multi-target learning for classification processing, outputting the classification result).
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over () in view of Kim (KR 102100699 B1) in view of Achmad (CN 111247595 A), in view of Lu (CN 110838108 A), and further in view of Park (PGPUB: 20200178918 A1).
Regarding claim 8. The combination teaches the method of claim 7 wherein inputting (42) the cognitive test information comprises inputting (42) to a pooling layer of the output portion (see page 15, lines 18-21, the convolutional neural network may include a pooling layer in which a pooling operation is performed in addition to a convolutional layer in which a convolution operation is performed. The pooling technique is a technique used to reduce the spatial size of data in the pooling layer).
However, the combination does not expressly teach that the output portion having a dense layer after the pooling layer.
Park teaches that the convolutional neural network is composed of a stack of convolutional modules that perform feature extraction. Each module consists of a convolutional layer followed by a pooling layer. The last convolutional module is followed by one or more dense layers that perform classification (see paragraph 43).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination by Park for providing Each module consists of a convolutional layer followed by a pooling layer. The last convolutional module is followed by one or more dense layers that perform classification, as the output portion having a dense layer after the pooling layer. Therefore, combining the elements from prior arts according to known methods and technique, such as The last convolutional module is followed by one or more dense layers that perform classification would yield predictable results.
Allowable Subject Matter
Claims 9-10 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.
Response to Arguments
Applicant's arguments filed 1/7/2026 have been fully considered but they are not persuasive.
In page 7, lines 19-23, applicant argues that None of the cited references teach or suggest "inputting the first type of image data into the neural network, the neural network having first and second input branches each including multiple layers, the first input branch being for the first type of image data and the second input branch being for a second type of image data" as recited by Claim 1.
Examiner respectfully disagrees. Kim teaches that the lesion integrated learning apparatus 10 may prepare a learning data set 200 (see FIG. 2) for performing learning of the above-described lesion area integrated learning model or severity integrated learning model. The learning data set 200 is an image of a lesion region obtained by extracting a region in which a lesion appears in the diagnostic images 201, 202, and 203 and the diagnostic images 201, 202, and 203 of the patient having different diseases. (211, 212, 213), and the type of disease corresponding to the lesion area (221a, 222a, 223a) and the severity of the disease (221b, 222b, 223b) indicating data (221, 222, 223) may be included have (see Fig. 2, page 6, lines 18-25); the lesion area learning unit 13 may perform learning on the integrated learning model 300 for the lesion area, where diagnostic images 201, 202, and 203 may be used as input. In particular, the diagnostic images 201, 202, and 203 input to the lesion area integrated learning model 300 may be normalized images through the image normalization unit 11 described above. In addition, the lesion area learning unit 13 may set and provide the target variable of the lesion area integrated learning model 300 as images of the lesion area 211, 212, and 213. Accordingly, the lesion area integrated learning model 300 may be trained to detect the images 211, 212, and 213 of the lesion area corresponding to the diagnostic images 201, 202, and 203, and furthermore, the lesion area integrated learning model. The 300 may be learned based on a convolutional neural network (CNN) technique or a pooling technique (see Fig. 3, page 6, lines 32-35 and 1-7).
As Kim taught that the different type of data are provided, as three input branches to model 300 of the neural network. Therefore, Kim clearly teaches the limitation of “inputting the first type of image data into the neural network, the neural network having first and second input branches each including multiple layers, the first input branch being for the first type of image data and the second input branch being for a second type of image data" as recited by Claim 1.
In page 8, lines 7-9, applicant argues that Kim does not disclose a single neural network that simultaneously includes multiple input branches. In other words, Kim does not disclose or suggest a neural network having first and second input branches.
Examiner respectfully disagrees. At first, the term “simultaneously includes multiple input branches” is not recited in the claim 1. In fact, Kim indeed teaches the three input branches shown in Figure 3, as a neural network having first and second input branches.
In page 9, lines 3-8, applicant argues that the lesion area integrated learning model 300 of Kim is best understood as a standalone neural network trained to generate lesion region images from diagnostic images, operating with a fixed input structure and producing image-based outputs for use in subsequent learning stages, rather than as a branched neural network or a model trained to output patient-level Alzheimer's disease classifications under varying image input conditions.
Examiner respectfully disagrees. At first, what kind of the neural network recited in claim 1 is not clearly specified, examiner interprets it as a general neural network. Therefore, the neural network taught by Kim teaches the neural network recited in claim 1. And further, the three inputs branches shown in Fig. 3 as a branched neural network.
Achmad teaches that classifying a subject as medical system with mild cognitive impairment (MCI) or Alzheimer ' s disease (AD), the system comprising: a computer system, learning machine one or more subjects the computer system is arranged to receive the unknown classification MCI or AD of the medical data and at least partially is configured to provide second the trained learning machine of the first trained first indication and providing the second indication, and combined to provide the first and second indication which subject with MCI and which subject has AD identification; a display, said display for displaying classification. computer system can be arranged to subject to classification is MCI provides MCI is further classification of the MCI early or advanced stage MCI (see page 11, lines 5-14). Achmad teaches to display (as output) the classification such as Alzheimer ' s disease.
In page 9, lines 9-13, applicant argues that None of the cited references teach or suggest "the neural network having an output portion having been trained to output the classification in response to the inputting of the first type of image data, the second type of image data, and both the first and the second type of image data" as recited by Claim 1.
Examiner respectfully disagrees. Kim teach input images and Achmad teaches output classification, as mentioned above. And further, Lu teaches that S73, the first image input to be predicted based on multi-target learning depth neural network feature extraction sub-network to down-sampling operation, extracting the first characteristic information of a prediction image, and the depth of the classifier neural network extracting feature information of the input based on the multi-target learning for classification processing, outputting the classification result (see Fig. 7, page 13, lines 5-9).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the Kim by Achmad and Lu to provide "the neural network having an output portion having been trained to output the classification in response to the inputting of the first type of image data, the second type of image data, and both the first and the second type of image data".
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to XIN JIA whose telephone number is (571)270-5536. The examiner can normally be reached 9:00 am-7:30pm.
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, Gregory Morse can be reached at (571)272-3838. 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.
/XIN JIA/Primary Examiner, Art Unit 2663