System and method for visualizing analysis result of endoscopic image

A method of visualizing an analysis result of an endoscopic image, the method comprising steps of:

• a) loading, by a model loading/condition setting unit, a plurality of analysis models and setting analysis conditions of the analysis models;
• b) determining, by an image analyzer, whether a situation requires image analysis, and reading an image frame and analyzing the image using the plurality of analysis models when the situation requires image analysis;
• c) determining, by an AM extractor/generator, whether an image analysis result by the image analyzer satisfies an analysis condition set by the model loading/condition setting unit;
• d) extracting, by the AM extractor/generator, AMs of the plurality of analysis models when the analysis condition is satisfied in the determining of the step c) and generating a CAM based on the extracted AMs;
• e) determining, by the AM extractor/generator, whether an analysis target is detected in a previous image frame and generating an ACAM by applying weights to CAMs extracted in the previous image frame and a current image frame when the analysis target is detected;
• f) extracting, by an activation value extractor, a peak activation value from the ACAM generated by the AM extractor/generator;
• and
• g) determining, by a closed curve extractor, whether the extracted peak activation value is greater than or equal to a detection confirmation threshold, extracting a closed curve of an AM when the extracted peak activation value is greater than or equal to the detection confirmation threshold, and drawing the closed curve on a corresponding analysis image.

A method of visualizing an analysis result of an endoscopic image, the method comprising steps of:

• a) loading, by a model loading/condition setting unit, a plurality of analysis models and setting analysis conditions of the analysis models;
• b) determining, by an image analyzer, whether a situation requires image analysis, and reading an image frame when the situation requires image analysis;
• c) determining, by the image analyzer, whether a mode is a detection mode or a classification mode, analyzing the image in the detection mode when the mode is the detection mode, and analyzing the image in the classification mode when the mode is the classification mode;
• d) determining, by an AM extractor/generator, whether an image analysis result by the image analyzer satisfies an analysis condition set by the model loading/condition setting unit;
• e) extracting, by the AM extractor/generator, AMs of the plurality of analysis models when the analysis condition is satisfied in the determining of the step d) and generating a CAM based on the extracted AMs;
• f) determining, by the AM extractor/generator, whether an analysis target is detected in a previous image frame and generating an ACAM by applying weights to CAMs extracted in the previous image frame and a current image frame when the analysis target is detected;
• g) extracting, by an activation value extractor, a peak activation value from the ACAM generated by the AM extractor/generator;
• h) determining, by a closed curve extractor, whether the extracted peak activation value is greater than or equal to a detection confirmation threshold, extracting a closed curve of an AM when the extracted peak activation value is greater than or equal to the detection confirmation threshold, and drawing the closed curve on a corresponding analysis image;
• and
• i) determining, by the image analyzer, whether to change an analysis mode, and changing the analysis mode to the classification mode when a current analysis mode is the detection mode and changing the analysis mode to the detection mode when the current analysis mode is the classification mode in a case of changing the analysis mode.