# US 8,787,636 B2 · Diagnostic imaging support in which image data of high magnification is generated to image data of low magnification for classification thereof > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 8,787,636 B2 - Document ID: US8787636B2 - Status: Granted - Kind: B2 - Application number: 13/148,902 - Filed: 2009-12-25 - Published: 2012-01-05 - Granted: 2014-07-22 - Earliest claimed: 2009-03-04 - Assignees: NEC Corporation - Applicants: Atsushi Marugame - Inventors: Atsushi Marugame - Web page: https://patentgrep.com/patent/US8787636B2 ## CPC - G06V 20/698: {Matching; Classification} - G06T 7/0012: {Biomedical image inspection} - G06T 2207/10056: Microscopic image - G06T 2207/20016: Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform - G06T 2207/30024: Cell structures in vitro; Tissue sections in vitro - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A diagnostic imaging support device includes an image data acquisition unit that acquires image data of high magnification of sample tissue, an image classifying unit that generates image data of low magnification from the image data of high magnification acquired by the image data acquisition unit, and classifies the generated image data of low magnification into group by each image data pattern of a plurality of pathological tissue, and an image evaluating unit that evaluates whether or not the image data of high magnification, which is to be a base of the image data of low magnification classified by the image classifying unit, is the pathological tissue of the classified group. ## Claims ### Claim 1 A diagnostic imaging support device comprising: image data acquisition portion that acquires image data of high magnification of sample tissue; image classifying portion that generates image data of low magnification from the image data of high magnification acquired by the image data acquisition portion, extracts a region of interest from the generated image data of low magnification, calculates a feature value indicating a feature of the region of interest, and classifies the generated image data of low magnification into group by each image data pattern of a plurality of pathological tissue according to the calculated feature value; and image evaluating portion that extracts a region of interest from the image data of high magnification based on the image data of high magnification which is a base of the image data of low magnification classified by the image classifying portion, and the region of interest of the image data of low magnification, calculates a feature value of the region of interest, and evaluates whether or not the image data of high magnification is pathological tissue of the classified group based on the calculated feature value. ### Claim 2 (depends on 1) The diagnostic imaging support device according to claim 1, wherein the image classifying portion calculates a total of a score function for each of the group according to the feature value and the score function, and classifies the image data of low magnification into the group with the largest total of the score function. ### Claim 3 (depends on 1) The diagnostic imaging support device according to claim 1, wherein the image classifying portion classifies the image data of low magnification according to probability distribution of the feature value. ### Claim 4 (depends on 1) The diagnostic imaging support device according to claim 1, wherein the image evaluating portion extracts the region of interest from the image data of high magnification, calculates a feature value indicating a feature of the region of interest, and performs the evaluation according to the calculated feature value. ### Claim 5 (depends on 4) The diagnostic imaging support device according to claim 4, wherein the image evaluating portion includes a learning mode that learns a relationship between the feature value and the evaluation result to set a learning parameter, and an execution mode that performs the evaluation according to the learning parameter. ### Claim 6 (depends on 5) The diagnostic imaging support device according to claim 5, wherein the image evaluating portion executes the learning mode again after executing the execution mode. ### Claim 7 (depends on 5) The diagnostic imaging support device according to claim 5, further comprising a learning database that stores initialization data for the learning by the image evaluating portion, and the learning parameter. ### Claim 8 (depends on 1) The image diagnostic support device according to claim 1, wherein the group by each of the image data pattern of the plurality of pathological tissue includes at least one of a normal, a benign tumor, a precancerous state, and a cancer state. ### Claim 9 (depends on 1) The diagnostic imaging support device according to claim 1, wherein the feature value is an average value or a variation value of at least mass density, density of a mass area, the mass area, a mass thickness, and a mass length per unit area in a cell nucleus. ### Claim 10 (depends on 1) The diagnostic imaging support device according to claim 1, wherein the region of interest is a region in which cell nucleus density is greater than or equal to a predetermined value. ### Claim 11 A diagnostic imaging support method comprising: acquiring image data of high magnification of sample tissue; generating image data of low magnification from the acquired image data of high magnification, extracting a region of interest from the generated image data of low magnification, calculating a feature value indicating a feature of the region of interest, and classifying the generated image data of low magnification into group by each image data pattern of a plurality of sample tissue according to the calculated feature value; and extracting a region of interest from the image data of high magnification based on the image data of high magnification which is a base of the image data of low magnification classified by the image classifying portion, and the region of interest of the image data of low magnification, calculating a feature value of the region of interest, and evaluating whether or not the image data of high magnification is pathological tissue of the classified group based on the calculated feature value. ### Claim 12 (depends on 11) The diagnostic imaging support method according to claim 11, further comprising calculating a total of a score function by each of the group according to the feature value and the score function, and classifying the image data of low magnification into the group with the largest total of the score function. ### Claim 13 (depends on 11) The diagnostic imaging support method according to claim 11, further comprising classifying the image data of low magnification according to probability distribution of the feature value. ### Claim 14 (depends on 11) The diagnostic imaging support method according to claim 11, further comprising extracting the region of interest from the image of high magnification, calculating the feature value indicating the feature of the region of interest, and performing the evaluation according to the calculated feature value. ### Claim 15 A non-transitory storage medium storing a program causing a computer to execute: a process that acquires image data of high magnification of sample tissue; a process that generates image data of low magnification from the acquired image data of high magnification, extracts a region of interest from the generated image data of low magnification, calculates a feature value indicating a feature of the region of interest, and classifies the generated image data of low magnification into a group by each image data pattern of a plurality of pathological tissue according to the calculated feature value; and a process that extracts a region of interested from the image data of high magnification based on the image data of high magnification which is a base of the classified image data of low magnification, and the region of interest of the image data of low magnification, calculates a feature value of the region of interest, and evaluates whether or not the image data of high magnification is pathological tissue of the classified group based on the calculated feature value.