Gastroscope image analysis system and method based on repair and selective enhancement, and equipment

Abstract

The invention belongs to the field of image recognition, particularly relates to a gastroscope image analysis system and method based on restoration and selective enhancement, and equipment, and aims to solve the problem that an existing gastroscope image recognition system cannot accurately recognize an early gastric cancer image. The method comprises the following steps: acquiring a narrow-band imaged gastroscope image as a to-be-detected image; performing preprocessing to obtain a to-be-detected image only containing gastric mucosa; obtaining a to-be-detected image without reflection through the reflection processing module; generating a composite image through the trained generative adversarial network, and automatically selecting a more vivid image; and enabling the trained gastroscope image recognition network to obtain the early gastric cancer probability of the image to be detected and obtain a region image of suspected early gastric cancer through a gradient weighted class activation mapping method. According to the method, selective enhancement of data features is carried out through the generative adversarial network, feature information most related to the classification task is automatically learned through the recognition model, and the accuracy of gastroscope image analysis is improved.

Description

technical field [0001] The invention belongs to the field of image recognition, and in particular relates to a gastroscope image analysis system, method and equipment based on restoration and selective enhancement. Background technique [0002] Gastric cancer (GC) is one of the most common malignancies in the world, with an estimated 1 million new cases each year. The prognosis of advanced gastric cancer is poor, while early gastric cancer (EGC) has a high rate of radical endoscopic resection due to less lymph node metastasis and a 5-year survival rate of over 90%. Therefore, timely and accurate identification of EGC is of great significance. However, the identification of EGCs is quite challenging. The average detection rate of EGC in my country is generally about 2%-5%, and the missed detection rate of EGC during gastroscopy is about 10% [3]. Considering the high incidence of gastric cancer, the number of missed diagnoses of EGC is overwhelming. [0003] To overcome thi...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the above problems in the prior art, that is, the existing gastroscope image analysis technology cannot accurately and quickly identify the gastroscope image, the present invention provides a gastroscope image a...

Images