Method and system for reading capsule endoscopy image based on artificial intelligence

摘要

The present invention relates to an artificial intelligence-based capsule endoscope image reading method and system, wherein the artificial intelligence-based capsule endoscope image reading system includes: a preprocessor for preprocessing an image (capsule endoscope image) captured by the capsule endoscope; And a convolutional neural network (CNN) that classifies an image with a lesion and an image without a lesion by taking the preprocessed capsule endoscope image as an input, wherein the convolutional neural network (CNN) receives the preprocessed capsule endoscope image as an input. layer; A convolutional layer for extracting features from the preprocessed capsule endoscope image input through the input unit; A maximum pooling layer that subsamples the features of the extracted small intestine endoscopy image to increase system stability and efficiency; A global average pooling layer that replaces the fully connected layer and obtains a class activation map (CAM); And an output layer for outputting a probability value of the presence of a lesion and a probability value of absence of a lesion for each of the capsule endoscopy images. According to the present invention, it is possible to reduce the reading time of a doctor who reads a large amount of endoscopic images taken with a capsule endoscope and increase the accuracy of reading, thereby enabling high-quality medical treatment.