Auxiliary diagnosis system and inspection device for early digestive tract cancer based on deep learning

Abstract

The present invention provides a system and device for auxiliary examination and diagnosis of digestive tract early cancer based on deep learning. The system includes a feature extraction network, an image classification model, an endoscope classifier and an early cancer recognition model; the feature extraction network is used to The model performs preliminary feature extraction on endoscopic images; the image classification model is used to extract preliminary features to obtain image classification features; the endoscope classifier is used to perform feature extraction on preliminary features to obtain endoscopic classification features and classify The image is classified; the early cancer recognition model is used to splicing the preliminary features, endoscopic classification features and image classification features to obtain the probability of early cancer lesions in the white light image, electronic staining image or chemical staining image of the corresponding part, or to obtain the corresponding part flushing reminder or location recognition reminder. The invention improves the quality of AI-assisted diagnosis and the diagnostic efficiency of digestive endoscopy.

Description

technical field [0001] The invention relates to medical inspection equipment, in particular to an auxiliary diagnosis system and inspection device for early digestive tract cancer. Background technique [0002] With the development of artificial intelligence technology based on deep learning, the application of artificial intelligence in the field of medical imaging diagnosis has attracted more and more attention. Through artificial intelligence technology, it can automatically judge possible lesions based on medical images, and complete automatic screening of medical images. At present, artificial intelligence technology has been widely studied in various fields such as breast cancer pathological examination, lung cancer detection, and cardiovascular imaging. [0003] Gastrointestinal diseases are frequently-occurring and common diseases, which seriously threaten people's life and health. Digestive endoscopy and chromoendoscopy are the first choice for diagnosing gastroin...

AI Technical Summary

Problems solved by technology

[0008] At present, in order to realize the screening of early digestive tract cancers through the diagnosis of digestive endoscopy in our country, we still face the following main problems: large hospitals have a large number of check-ups, and the doctors are tired of work; small hospitals have a small number of check-ups, and the work of doctors is extremely unsaturated ; Doctors have a long training period, and the improvement of their diagnostic level is slow; during the inspection process, they often encounter blurred vision, weak cleaning methods, time-consuming and poor results; dyeing solutions used in chromoendoscopy are not complete, and doctors' self-made methods vary greatly, and the concentration used is not uniform , the methods of use vary greatly, and it is impossible to form a standardized diagnostic map

Images