Intestinal neuron neuron dysplasia recognition method based on Swinin-Unet algorithm

Abstract

The invention discloses a pathological recognition method for intestinal neuron dysplasia based on a Swin-Unet algorithm, and the method comprises the following steps: obtaining submucosal and intermuscular nerve plexus in an H&E dyeing image of an intestinal tissue hematoxylin-eosin dyeing section, collecting a ganglion cell image in the nerve plexus, and carrying out the preprocessing of the ganglion cell image to serve as a training data set; performing data enhancement on the training data set; constructing an image segmentation model; training the image segmentation model by using the training data set; constructing an image classification model; preprocessing the training data set again; training the image classification model by using the training data set; and segmenting and classifying an image to be detected by using the trained image segmentation model and image classification model to realize the judgment of the development condition of the intestinal ganglion. According to the invention, accurate and stable identification and classification of the ganglion cells in submucosa and intermuscular nerve plexus in the intestinal tissue slice dyeing image can be realized, so that subsequent diagnosis is assisted.

Description

technical field [0001] The invention relates to the technical field of deep learning, in particular to a method for identifying abnormal development of enteric neurons based on the Swin-Unet algorithm. Background technique [0002] Enteric neuronal dysplasia mainly includes intestinal aganglionosis and its homologous diseases. Intestinal aganglionosis (Aganglionosis; Hirschsprung Disease, HSCR) is due to the absence of ganglion cells in the distal intestinal canal of the digestive tract, resulting in the disappearance of recto-anal inhibition reflex. A group of disorders characterized by obstruction. Certain manifestations similar to intestinal aganglionosis, such as abdominal distension, bowel dilation, chronic constipation, but enteric ganglion cells are present in the rectum, known as Hirschsprung disease (Allied Disorders of Hirschsprung Disease), also known as enteric nerve Abnormal development. According to the latest guidelines, Hirschsprung homologous disease is i...

AI Technical Summary

Problems solved by technology

However, despite the impressive performance achieved by CNN, it cannot learn global and long-range semantic information interactions well due to the limitations of convolution operations, often showing limitations in explicitly modeling long-range dependencies.

Images