Gastroscope image intelligent processing method and device

Abstract

The invention provides a gastroscope image intelligent processing method and a gastroscope image intelligent processing device. The gastroscope image intelligent processing method comprises the steps of: dynamically acquiring a first gastroscope image; segmenting a segmented region having lesion characteristics from the first gastroscope image, labeling primary characteristic information and position information corresponding to the segmented region on the first gastroscope image to serve as second gastroscope images, and outputting the second gastroscope images; receiving the second gastroscope images, performing combinatory analysis on the primary characteristic information and the position information in a plurality of the second gastroscope images to form secondary characteristic information and regional position information corresponding to the secondary characteristic information, and outputting a third gastroscope image labeled with the secondary characteristic information and the regional position information; and receiving the third gastroscope image and displaying the same. The gastroscope image intelligent processing method and the gastroscope image intelligent processing device can assist the doctor in diagnosis, obtain a pathological suggested region, and significantly reduce the probability of missed diagnosis and misdiagnosis.

Description

technical field [0001] The present application relates to the field of gastroscope image processing, in particular to a method and device for intelligent processing of gastroscope images. Background technique [0002] At present, gastroscopy is still the most direct, accurate and reliable diagnostic method for gastric cancer. The precancerous lesions of gastric cancer often involve extensive areas of gastric mucosa. Routine endoscopic biopsy adopts segmented and random multi-point sampling methods, and the possibility of missing focal cancerous tissues is very high, especially for inexperienced endoscopic doctors. "Cancer is invisible"; secondly, the diagnosis of atypical hyperplasia and early gastric cancer under the microscope requires a large amount of case accumulation and training. Diagnosing the same group of gastroscopy biopsy specimens between different pathologists or the same pathologist multiple times is difficult. There are certain differences. [0003] However...

AI Technical Summary

Problems solved by technology

The precancerous lesions of gastric cancer often involve extensive areas of gastric mucosa. Routine endoscopic biopsy adopts segmented and random multi-point sampling methods, and the possibility of missing focal cancerous tissues is very high, especially for inexperienced endoscopic doctors. "Cancer is invisible"; secondly, the diagnosis of atypical hyperplasia and early gastric cancer under the microscope requires a large amount of case accumulation and training. Diagnosing the same group of gastroscopy biopsy specimens between different pathologists or the same pathologist multiple times is difficult. there are certain differences

[0003] However, traditional gastroscopy has the following disadvantages: 1. The diagnosis rate of early gastric cancer is low, and the rate of missed detection and misdiagnosis are high

2. The existing gastroscopy system only provides images to doctors, but does not perform preliminary screening, feature extraction and labeling of images, and doctors have to face a large number of medical images, which increases the workload of doctors and is inefficient And it is very easy to cause doctors to "miss diagnosis"

3. Although the lesion can be highlighted and visualized, it is not targeted at the cancer tissue, so the judgment of the nature of the lesion still depends on the experience of the operating physician

Images