Intestinal tumor microscopic hyperspectral image processing method based on convolutional neural network

Abstract

The invention discloses an intestinal tumor microscopic hyperspectral image processing method based on a convolutional neural network, and particularly relates to the technical field of gastrointestinal tumor detection, and the hyperspectral image processing method comprises the following steps: 1, obtaining a hyperspectral initial image, and irradiating the gastrointestinal tumor position of a patient by using related equipment. According to the invention, the hyperspectral initial image is segmented; a second sub-image is obtained by using the first sub-image, the feature points in each section of the second sub-image is extracted in a convolutional neural network manner; a third sub-image is constructed, and the feature points are integrated in each group of third sub-images; therefore, the hyperspectral target image formed by combining the feature points is obtained, the hyperspectral initial image can be effectively processed, the hyperspectral initial image is more visual and concise, the processing efficiency of a large amount of data in the hyperspectral initial image is improved, and meanwhile the disease diagnosis accuracy of people is improved.

Description

technical field [0001] The invention relates to the technical field of gastrointestinal tumor detection, and more specifically, the invention relates to a method for processing microscopic hyperspectral images of intestinal tumors based on a convolutional neural network. Background technique [0002] Gastrointestinal tumors, benign and malignant tumors that occur in the small intestine and large intestine. Clinical manifestations vary according to the nature and location of tumor occurrence. In general, benign tumors can be asymptomatic or have very mild symptoms. Some malignant tumors have no obvious symptoms in the early stage, which affects diagnosis, treatment and prognosis. Among intestinal tumors, the incidence of tumors in the small intestine is lower than those in the esophagus, stomach, and large intestine. The clinical manifestations of patients are anemia, weight loss, increased stool frequency, deformation, and mucus and bloody stools. Sometimes an abdominal ...

AI Technical Summary

Problems solved by technology

[0006] However, when the above-mentioned technical solution is actually used, there are still many shortcomings. For example, the hyperspectral image in the above-mentioned solution is analyzed in batches. As the number of bands in the hyperspectral image increases, the internal data The amount increases exponentially, resulting in an increase in the amount of image analysis during image processing, requiring a chip capable of processing higher-resolution images for analysis and processing, resulting in increased computing costs, which is not convenient for practical applications