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 .
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 06/29/2026 has been entered.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-21 are rejected under 35 U.S.C. 102(a1) as being anticipated by Ohno (US 2023/0230248 A1).
Regarding claim 1, Ohno discloses an information processing system (e.g., FIG. 1 is a diagram showing the overall configuration of an information processing system including an information processing apparatus according to the present embodiment, paragraph 25, figure 1), comprising:
a central processing unit (CPU) (e.g., the control unit 117 is implemented by a CPU (Central Processing Unit), a GPU (Graphical Processing Unit), or the like, paragraph 32) configured to:
generate a learned model based on learning data, wherein the learned model estimates a health condition of one of a patient or a subject (e.g., a CT image of the chest including the lesion of a subject is first obtained by a CT apparatus, and then a diagnosis result is estimated by a learning model using the obtained CT image as input data, paragraph 22);
acquire adjustment information based on a feature value of the learning data (e.g., In the present embodiment, a configuration using a learning model for estimating the severity (disease stage) of a disease is described as an example, but a configuration for estimating any or all of the information can be used. The learning model can also perform iterative learning based on training data including input data and labels, paragraph 35);
process measurement data of a biological sample based on the acquired adjustment information, wherein the biological sample is a target for estimation of a diagnosis result (e.g., The obtaining unit 118 reads and obtains from the data server 102 a medical image of a subject captured by the medical imaging apparatus 101 and information associated with the medical image. The information associated with the medical image may include subject information such as a subject ID, subject height, weight, age, gender, body fat, blood pressure, pregnancy status, heart rate, body temperature, etc. It may be examination information such as a condition, an imaging region, an imaging date and time, or an imaging location. The obtaining unit 118 may obtain all of the information associated with the medical images stored in the data server 102, or may obtain only some items, paragraph 34);
input the processed measurement data to the learned model; and estimate the diagnosis result based on the input of the processed measurement data to the learned model (e.g., The estimation unit 119 estimates a diagnosis result from the medical image of the subject obtained by the obtaining unit 118. In the present embodiment, a learning model that has performed deep learning in advance is used to estimate the diagnosis result from the medical image of the subject. The learning model, which will be described in detail later, is constructed, for example, by performing supervised learning with a neural network, using pairs of input data and labels as training data…, paragraph 35).
Regarding claim 2, Ohno discloses wherein the CPU is further configured to:
generate the learned model based on a learning data set, wherein the learning data set includes a plurality of pieces of the learning data (e.g., FIG. 4A is a diagram illustrating an example of a process of generating a learning model according to the first embodiment, paragraph 16);
acquire the adjustment information based on statistical information on a feature value of the plurality of pieces of the learning data (e.g., In S401, the learning unit included in the control unit 117 obtains medical images and labels that serve as training data from the data server 102. Note that the medical images and labels do not necessarily have to be obtained from the data server 102, and may be obtained from another data server. Here, in the present embodiment, the labels are identification information identifying, for example, the presence or absence of disease, the severity of the disease (stage), the type of disease, the presence or absence of metastasis, the location of the metastasis, the location of the tumor, the size of the tumor, the number of tumors, or the like, paragraph 48); and
adjust a first feature value of the measurement data based on the acquired adjustment information (e.g., the learning unit provides the set of input data and labels included in the training data to a neural network 402 configured by combining perceptions, and performs forward propagation such that the weighting for each perceptron in the neural network 402 is changed such that the output of the neural network 402 becomes equal to the label, paragraph 51).
Regarding claim 3, Ohno discloses wherein the CPU is further configured to adjust the first form training data, paragraph 59),
the first form training data, and a learning model is generated that outputs the type of disease and the presence or absence of metastasis, paragraph 59), and
the second feature value of the measurement data falls within the specific range for the statistical information (e.g., Alternatively, different learning models may be generated for each label associated with input data. For example, a first learning model is generated from training data in which the type of disease is labelled for each medical image, and a second learning model is generated from the training data in which the presence or absence of metastasis is labelled for each medical image, paragraph 60).
Regarding claim 4, Ohno discloses wherein the specific range is one of:
a first range in which reliability of the diagnosis result is one of equal to or greater than a specific value; or a second range in which the reliability of the diagnosis result is expected to be one of equal to or greater than the specific value (e.g., Alternatively, a plurality of learning models may be generated by associating the same label with different input data. For example, in S401, CT images and MRI images are obtained. Then, a first learning model is generated from training data configured such that the severity (the disease stage) of a disease is labelled for each obtained CT image, and furthermore, a second learning model is generated from training data configured such that the severity (the disease stage) of the disease is labelled for each MRI image, paragraph 61).
Regarding claim 5, Ohno discloses wherein
the CPU processing unit is further configured to adjust the first feature value of the measurement data based on one of a median value, an average value, or a centroid value in the statistical information (e.g., Furthermore, the learning unit may combine a plurality of labels for one piece of input data to form training data, paragraph 59),
the first form training data, and a learning model is generated that outputs the type of disease and the presence or absence of metastasis, paragraph 59), and
the second feature value of the measurement data is approximated to the one of the median value, the average value, or the centroid value in the statistical information (e.g., Alternatively, different learning models may be generated for each label associated with input data. For example, a first learning model is generated from training data in which the type of disease is labelled for each medical image, and a second learning model is generated from the training data in which the presence or absence of metastasis is labelled for each medical image, paragraph 60).
Regarding claim 6, Ohno discloses wherein the CPU is further configured to:
generate a conversion formula based on the statistical information, wherein the first
adjust the first
Regarding claim 7, Ohno discloses wherein
the CPU is further configured to output the adjusted first feature value of the measurement data based on a neural network, wherein the first
the second feature value of the measurement data is approximated to the feature value of the plurality of pieces of the learning data, and the neural network receives image data as an input (e.g., the learning unit may learn all the medical images obtained in S401, or may learn only some of the obtained medical images. Furthermore, the learning unit may use divided images obtained by dividing an obtained medical image into a plurality of regions as the input data, or may extract only a partial region of interest from the obtained medical image and use the extracted part as the input data, paragraph 58); and
estimate the diagnosis result based on input of the second feature value of the measurement data to the learned model (e.g., different learning models may be generated for each label associated with input data. For example, a first learning model is generated from training data in which the type of disease is labelled for each medical image, and a second learning model is generated from the training data in which the presence or absence of metastasis is labelled for each medical image, paragraph 60).
Regarding claim 8, Ohno discloses wherein the CPU is further configured to adjust a weight parameter of the neural network based on the statistical information on the feature value of the plurality of pieces of the learning data (e.g., in the convolutional layer, a filter with predetermined parameters is applied to the input image data to perform feature extraction such as edge extraction. The predetermined parameters in this filter correspond to the weights of the neural network, which are learned by repeating the forward and back propagation described above, paragraph 54).
Regarding claim 9, Ohno discloses wherein the CPU is further configured to:
evaluate the diagnosis result; and adjust the weight parameter of the neural network based on the evaluation of the diagnosis result (e.g., the learning unit adjusts the weighting values so as to reduce the error in the output of each perceptron by a method called back propagation. More specifically, the learning unit calculates the error between the output of the neural network 402 and the label, and modifies the weighting values so as to reduce the calculated error, paragraph 52).
Regarding claim 10, Ohno discloses wherein
the CPU is further configured to select one of a plurality of neural networks based on a respective evaluation output for each of a plurality of diagnosis results (e.g., The learning model, which will be described in detail later, is constructed, for example, by performing supervised learning with a neural network, using pairs of input data and labels as training data. In the present embodiment, the diagnosis result indicates, paragraph 35),
the plurality of neural networks includes the neural network (e.g., Furthermore, the specific algorithm for generating the learning model is not limited to the above, and in addition to deep learning using neural networks, for example, support vector machines, Bayesian networks, or random forests may be used, paragraph 35),
a first neural network of the plurality of neural networks is different from a second neural network of the plurality of neural networks, and the plurality of diagnosis results includes the diagnosis result (e.g., the neural network 402 has a structure in which a large number of processing units 403 are arbitrarily connected. Examples of processing units 403 include processing units for a convolution operation, normalization processing such as batch normalization, or processing using activation functions such as ReLU, Sigmoid, and Softmax, each having a set of parameters to describe the processing, paragraph 53).
Regarding claim 11, Ohno discloses further comprising an interface configured to:
present a respective evaluation output for each of a plurality of diagnosis results, and receive a user input based on the presented respective evaluation output (e.g., display screen 501 displays a list of information about treatment methods performed in past cases identified from diagnosis results estimated from a medical image of a subject, paragraph 79), wherein the CPU is further configured to select one of a plurality of neural networks based on the user input,
the plurality of neural networks includes the neural network, and the plurality of diagnosis results includes the diagnosis result (e.g., in the convolutional layer, a filter with predetermined parameters is applied to the input image data to perform feature extraction such as edge extraction. The predetermined parameters in this filter correspond to the weights of the neural network, which are learned by repeating the forward and back propagation described above, paragraph 54).
Regarding claim 12, Ohno discloses wherein the CPU is further configured to:
identify a piece of feature value from the first feature value of the measurement data, wherein the identified piece of feature value adversely affects the estimation of the diagnosis result, and adjust the piece of feature value (e.g., In the fully-connected layer, image data whose features have been extracted through the convolutional layer and the pooling layer are connected to one node, and a value obtained by the conversion using the activation function is output. Here, the activation function is a function that sets all output values less than 0 to 0, and is used to send only the outputs equal to or greater than a certain threshold to the output layer as meaningful information, paragraph 56).
Regarding claim 13, Ohno discloses further comprising an interface configured to:
present a relationship between the statistical information and a set of feature values of the measurement data (e.g., Various statistical values (average, median, maximum, etc.) based on the results evaluated in past cases may be used as evaluation values for the respective indicators. Different statistical values may be used for each indicator or the same statistical value may be used for all indicators, paragraph 83); and
receive an adjustment value of a second feature value of the measurement data based on a user input, wherein the set of feature values of the measurement data includes the first feature value of the measurement data, the CPU is further configured to adjust the set of feature values of the measurement data to a third feature value of the measurement data, and the third feature value of the measurement data is the adjustment value of the second feature value of the measurement data (e.g., In the fully-connected layer, image data whose features have been extracted through the convolutional layer and the pooling layer are connected to one node, and a value obtained by the conversion using the activation function is output. Here, the activation function is a function that sets all output values less than 0 to 0, and is used to send only the outputs equal to or greater than a certain threshold to the output layer as meaningful information, paragraphs 55, 56).
Regarding claim 14, Ohno discloses wherein the CPU is further configured to;
generate a plurality of learned models based on a plurality of pieces of the learning data, wherein the plurality of learned models includes the learned model (e.g., a learning model that has performed deep learning in advance is used to estimate the diagnosis result from the medical image of the subject, paragraph 35); and
select a specific learned model from the plurality of learned models, wherein
the specific learned model is generated based on a specific piece of the learning data of the plurality of pieces of the learning data (e.g., In the present embodiment, a configuration using a learning model for estimating the severity (disease stage) of a disease is described as an example, but a configuration for estimating any or all of the information can be used. The learning model can also perform iterative learning based on training data including input data and labels. The learning model may also be used to learn another model through transfer learning or fine-tuning, or further learning (additional learning) may be performed on the learning mode, paragraph 35), and
a feature value of the specific piece of the learning data is close to a learning model for estimating the diagnosis result may be generated by a learning unit (not shown) included in the information processing apparatus 103, or may be a model generated by an information processing apparatus other than the information processing apparatus 103. Furthermore, the specific algorithm for generating the learning model is not limited to the above, and in addition to deep learning using neural networks, for example, support vector machines, Bayesian networks, or random forests may be used, paragraph 35).
Regarding claim 15, Ohno discloses wherein the learning data and the measurement data are image data (e.g., a learning model that has performed deep learning in advance is used to estimate the diagnosis result from the medical image of the subject, paragraph 35), and
the
Regarding claim 16, Ohno discloses further including a display unit configured to present information on a user interface (e.g., figure 5), wherein
the 501 displays a list of information about treatment methods performed in past cases identified from diagnosis results estimated from a medical image of a subject, paragraph 79),
the CPU is further configured to specify a second physical condition at a time of acquisition of the measurement data (e.g., The above-described manner of displaying information on the display screen 501 is merely an example and is not limited thereto. For example, instead of displaying a list, the display control unit 143 may sequentially display candidate treatment methods that are most likely to be options for the user, paragraph 80),
the second physical condition is specified based on approximation of a feature value of the measurement data to the feature value of the learning data, and
the display unit is further configured to present the second physical condition on the user interface (e.g., Alternatively, the display control unit 143 may display the candidate treatment methods in the order from the standard treatment. Furthermore, the display control unit 143 may highlight the treatment methods that are frequently used as the treatment method. The control unit 141 may further have a search function of screening the extracted past cases based on the subject information and/or the like, paragraph 80).
Regarding claim 17, Ohno discloses wherein
the first physical condition is a parameter (e.g., Here it is assumed by way of example that the display screen 501 displays information on a diagnosis result for a case of a subject diagnosed as breast cancer with a severity corresponding to an “early” stage and information is given relating to past cases diagnosed as “early”-stage breast cancer including cases (95 cases)…,paragraph 79), and
the information processing system further comprises an interface configured to receive a user input to adjust the parameter in a process of acquisition of one of the learning data or the measurement data (e.g., the accepting unit 144 included in the information display apparatus 104 accepts an arbitrary selection from the user regarding the treatment methods displayed on the display screen 501, and changes the display screen according to the selection, paragraph 81).
Regarding claim 18, Ohno discloses wherein the learning data and the measurement data are medical images (e.g., a display screen 501 displays a list of information about treatment methods performed in past cases identified from diagnosis results estimated from a medical image of a subject, paragraph 79).
Regarding claim 19, Ohno discloses a non-transitory computer-readable medium having stored thereon, computer-executable instructions which, when executed by a computer, cause the computer to execute operations (e.g., a system or an apparatus via a network or a storage medium, and reading and executing the program by one or more processors of a computer in the system or the apparatus, paragraph 104), the operations comprising:
generating a learned model based on learning data, wherein the learned model estimates a health condition of one of a patient or a subject (e.g., a CT image of the chest including the lesion of a subject is first obtained by a CT apparatus, and then a diagnosis result is estimated by a learning model using the obtained CT image as input data, paragraph 22);
acquiring a feature value of the learning data (e.g., In the present embodiment, a configuration using a learning model for estimating the severity (disease stage) of a disease is described as an example, but a configuration for estimating any or all of the information can be used. The learning model can also perform iterative learning based on training data including input data and labels, paragraph 35);
outputting adjustment information based on a result of comparison between a feature value of measurement data of a biological sample and the acquired feature value of the learning data, wherein the biological sample is a target for estimation of a diagnosis result (e.g., The estimation unit 119 estimates a diagnosis result from the medical image of the subject obtained by the obtaining unit 118. In the present embodiment, a learning model that has performed deep learning in advance is used to estimate the diagnosis result from the medical image of the subject. The learning model, which will be described in detail later, is constructed, for example, by performing supervised learning with a neural network, using pairs of input data and labels as training data…, paragraph 35); and 90
processing the measurement data of the biological sample based on the adjustment information (e.g., The obtaining unit 118 reads and obtains from the data server 102 a medical image of a subject captured by the medical imaging apparatus 101 and information associated with the medical image. The information associated with the medical image may include subject information such as a subject ID, subject height, weight, age, gender, body fat, blood pressure, pregnancy status, heart rate, body temperature, etc. It may be examination information such as a condition, an imaging region, an imaging date and time, or an imaging location. The obtaining unit 118 may obtain all of the information associated with the medical images stored in the data server 102, or may obtain only some items, paragraph 34).
Regarding claim 20, claim 20 is a biological sample processing device with limitations similar of limitations of claim 1. Therefore, claim 20 is rejected as set forth above as claim 1.
Regarding claim 21, claim 21 is a non-transitory computer-readable medium claim with limitations similar of limitations of claim 1. Therefore, claim 21 is rejected as set forth above as claim 1.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to QUANG N VO whose telephone number is (571)270-1121. The examiner can normally be reached Monday-Friday, 7AM-4PM, EST.
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, Abderrahim Merouan can be reached at 571-270-5254. 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.
/QUANG N VO/Primary Examiner, Art Unit 2683
Prosecution Insights