# US 12,002,272 B2 · Method and device for classifing densities of cells, electronic device using method, and storage medium > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,002,272 B2 - Document ID: US12002272B2 - Status: Granted - Kind: B2 - Application number: 17/564,496 - Filed: 2021-12-29 - Published: 2022-06-30 - Granted: 2024-06-04 - Earliest claimed: 2020-12-30 - Assignees: Hon Hai Precision Industry Co., Ltd. - Applicants: Hon Hai Precision Industry Co., Ltd. - Inventors: Wan-Jhen Lee; Chih-Te Lu; Chin-Pin Kuo - Web page: https://patentgrep.com/patent/US12002272B2 ## CPC - G06V 20/693: {Acquisition} - G06N 3/045: Combinations of networks - G06N 3/0455: Auto-encoder networks; Encoder-decoder networks - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/08: Learning methods - G06N 3/09: Supervised learning - G06T 7/0012: {Biomedical image inspection} - G06V 10/454: {Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]} - G06V 10/82: using neural networks - G06V 20/698: {Matching; Classification} - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30024: Cell structures in vitro; Tissue sections in vitro - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A method for classifying cells densities by cell images being input into artificial computer intelligence obtains positional information of all central points of all groups of first encoding features generated when training a model of convolutional neural network and ranges of densities of images of biological cells represented by different central points. The method inputs a test image of the biological cells into a trained model of the convolutional neural network to encode the test image, to obtain a second encoding feature. The method also determines a central point nearest to the second encoding feature according to the positional information. The method determines a range of densities of the test image according to the ranges of densities of the images represented by different central points and the central point nearest to the second encoding feature. An electronic device and a non-transitory storage medium are also disclosed. ## Claims ### Claim 1 A method for classifying densities of cells comprising: obtaining positional information of all central points of all groups of first encoding features generated when training a model of convolutional neural network and ranges of densities of images of biological cells represented by different central points; each of the groups of the first encoding features comprising a central point and corresponding to a plurality of images of the biological cells within a range of densities; inputting a test image of the biological cells into a trained model of the convolutional neural network to encode the test image of the biological cells, to obtain a second encoding feature; determining a central point nearest to the second encoding feature according to the positional information of all the central points of all the groups of the first encoding features; determining a range of densities of the test image of the biological cells according to the ranges of densities of the images of the biological cells represented by different central points and the central point nearest to the second encoding feature. ### Claim 2 (depends on 1) The method according to claim 1, wherein the method further comprises: obtaining a plurality of training images of biological cells with different densities; inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain the trained model of the convolutional neural network. ### Claim 3 (depends on 2) The method according to claim 2, wherein the inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain the trained model of the convolutional neural network comprises: inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network; determining the central point of each of the groups of the first encoding features. ### Claim 4 (depends on 3) The method according to claim 3, wherein before the inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network, the method further comprises: dividing the training images of the biological cells with different densities into a plurality of training images of the biological cells with a plurality of different ranges of densities. ### Claim 5 (depends on 4) The method according to claim 4, wherein the inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network comprises: inputting the training image of the biological cells with one density into the model of the convolutional neural network to train the model of the convolutional neural network; obtaining the first encoding feature generated by encoding the training image of the biological cells with the one density when applying the training process; determining whether all the training images of the biological cells are input into the model of the convolutional neural network to train the model; dividing all the first encoding features into a plurality of different groups of the first encoding features according to the ranges of the densities of the training images of the biological cells if all the training images of the biological cells are input into the model of the convolutional neural network to train the model. ### Claim 6 (depends on 5) The method according to claim 5, wherein the inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network further comprises: continuously inputting the training image of the biological cells with another density into the model of the convolutional neural network to train the model of the convolutional neural network and obtaining the first encoding feature, until all the training images of the biological cells are input into the model of the convolutional neural network to train the model, if not all training images of the biological cells are input into the model of the convolutional neural network to train the model. ### Claim 7 (depends on 3) The method according to claim 3, wherein the determining the central point of each of the groups of the first encoding features comprises: calculating an average value of each group of the first encoding features to determine a central point of each group of the first encoding feature. ### Claim 8 An electronic device comprising: a storage device; at least one processor; and the storage device storing one or more programs, which when executed by the at least one processor, cause the at least one processor to: obtain positional information of all central points of all groups of first encoding features generated when training a model of convolutional neural network and ranges of densities of images of biological cells represented by different central points; each of the groups of the first encoding features comprising a central point and corresponding to a plurality of images of the biological cells within a range of densities; input a test image of the biological cells into a trained model of the convolutional neural network to encode the test image of the biological cells, to obtain a second encoding feature; determine a central point nearest to the second encoding feature according to the positional information of all the central points of all the groups of the first encoding features; determine a range of densities of the test image of the biological cells according to the ranges of densities of the images of the biological cells represented by different central points and the central point nearest to the second encoding feature. ### Claim 9 (depends on 8) The electronic device according to claim 8, further causing the at least one processor to: obtain a plurality of training images of biological cells with different densities; input the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain the trained model of the convolutional neural network. ### Claim 10 (depends on 9) The electronic device according to claim 9, further causing the at least one processor to: input the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network; determine the central point of each of the groups of the first encoding features. ### Claim 11 (depends on 10) The electronic device according to claim 10, before the inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network, further causing the at least one processor to: divide the training images of the biological cells with different densities into a plurality of training images of the biological cells with a plurality of different ranges of densities. ### Claim 12 (depends on 11) The electronic device according to claim 11, further causing the at least one processor to: input the training image of the biological cells with one density into the model of the convolutional neural network to train the model of the convolutional neural network; obtain the first encoding feature generated by encoding the training image of the biological cells with the one density when applying the training process; determine whether all the training images of the biological cells are input into the model of the convolutional neural network to train the model; divide all the first encoding features into a plurality of different groups of the first encoding features according to the ranges of the densities of the training images of the biological cells if all the training images of the biological cells are input into the model of the convolutional neural network to train the model. ### Claim 13 (depends on 12) The electronic device according to claim 12, further causing the at least one processor to: continuously input the training image of the biological cells with another density into the model of the convolutional neural network to train the model of the convolutional neural network and obtain the first encoding feature, until all the training images of the biological cells are input into the model of the convolutional neural network to train the model, if not all training images of the biological cells are input into the model of the convolutional neural network to train the model. ### Claim 14 (depends on 10) The electronic device according to claim 10, further causing the at least one processor to: calculate an average value of each group of the first encoding features to determine a central point of each group of the first encoding feature. ### Claim 15 A non-transitory storage medium storing a set of commands, when the commands being executed by at least one processor of an electronic device, causing the at least one processor to: obtain positional information of all central points of all groups of first encoding features generated when training a model of convolutional neural network and ranges of densities of images of biological cells represented by different central points; each of the groups of the first encoding features comprising a central point and corresponding to a plurality of images of the biological cells within a range of densities; input a test image of the biological cells into a trained model of the convolutional neural network to encode the test image of the biological cells, to obtain a second encoding feature; determine a central point nearest to the second encoding feature according to the positional information of all the central points of all the groups of the first encoding features; determine a range of densities of the test image of the biological cells according to the ranges of densities of the images of the biological cells represented by different central points and the central point nearest to the second encoding feature. ### Claim 16 (depends on 15) The non-transitory storage medium according to claim 15, further causing the at least one processor to: obtain a plurality of training images of biological cells with different densities; input the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain the trained model of the convolutional neural network. ### Claim 17 (depends on 16) The non-transitory storage medium according to claim 16, further causing the at least one processor to: input the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network; determine the central point of each of the groups of the first encoding features. ### Claim 18 (depends on 17) The non-transitory storage medium according to claim 17, before the inputting the training images of the biological cells with different densities one by one into the model of the convolutional neural network to train one by one to obtain different groups of the first encoding features of the trained images of the biological cells with different densities, until a training of the model of the convolutional neural network is completed, to obtain the trained model of the convolutional neural network, further causing the at least one processor to: divide the training images of the biological cells with different densities into a plurality of training images of the biological cells with a plurality of different ranges of densities. ### Claim 19 (depends on 18) The non-transitory storage medium according to claim 18, further causing the at least one processor to: input the training image of the biological cells with one density into the model of the convolutional neural network to train the model of the convolutional neural network; obtain the first encoding feature generated by encoding the training image of the biological cells with the one density when applying the training process; determine whether all the training images of the biological cells are input into the model of the convolutional neural network to train the model; divide all the first encoding features into a plurality of different groups of the first encoding features according to the ranges of the densities of the training images of the biological cells if all the training images of the biological cells are input into the model of the convolutional neural network to train the model. ### Claim 20 (depends on 17) The non-transitory storage medium according to claim 17, further causing the at least one processor to: calculate an average value of each group of the first encoding features to determine a central point of each group of the first encoding feature.