Method for segmenting epithelial tissue in esophageal pathological image

Abstract

The invention provides a method for segmenting an epithelial tissue in an esophageal pathological image. The method comprises the following steps: a) carrying out dyeing correction and gray level treatment; b) selecting training and testing samples; c) segmenting and labeling an image; d) constructing a convolution neural network model; e) processing a test image; f) acquiring a pre-heating image;g) processing the pre-heating image; h) calculating the accuracy and the recall rate. By adopting the epithelium segmentation method provided by the invention, the classification is carried out on the pixel level, particularly the segmentation of an epithelial boundary region. Therefore, the method has obvious advantages in segmentation precision. Effective characteristics and expressions can beautomatically learned by the method. The complicated manual characteristic selecting process is avoided, and the practical application requirements can be met. Epithelial tissues can be precisely segmented from images obtained by using different scanners from different hospitals. The method is an indispensable step for image processing in computer-aided diagnosis for constructing esophagus cancer.

Description

technical field [0001] The present invention relates to a method for segmenting epithelial tissue in pathological images of the esophagus, and more specifically, to a method for segmenting epithelial tissue in pathological images of the esophagus using a convolutional neural network model constructed from sample data. Background technique [0002] After examining a patient's biological tissue samples, the pathologist's report is often the gold standard for many diseases. With cancer in particular, a pathologist's diagnosis can have a profound impact on a patient's treatment. Pathology slide review is a very complex task that requires years of training to do well, as well as extensive expertise and experience. [0003] Esophageal cancer is a common malignant tumor in life, which seriously affects human health. The incidence of esophageal cancer in my country ranks among the highest in the world, and there are a large number of new cases of esophageal cancer every year. At ...

AI Technical Summary

Problems solved by technology

For a complete full-scan section of esophageal pathological tissue, its size is about 100,000×700,000 pixels, and it needs to occupy 1.5G of hard disk space to store on the computer. This high-resolution, large-scale image is very important for computer hardware and image analysis. Algorithms are very challenging

Images