Multi-organ segmentation method based on self-supervised feature small sample learning

Abstract

The invention discloses a multi-organ segmentation method based on self-supervised feature small sample learning, and mainly solves the problem of poor multi-organ segmentation effect by using a small sample learning segmentation method in the prior art. According to the scheme, the method comprises the steps of: using a superpixel segmentation method for generating a large amount of data containing pseudo labels from an initial data set, and selecting images and the pseudo labels from the data to serve as a support set; generating a query set by adopting a data enhancement method; extracting image features of the support set and the query set through a pre-trained encoder by using self-supervised feature learning, and then calculating the similarity of the image features to obtain foreground information and prior probability auxiliary information feature maps; constructing a segmentation network to carry out feature refining on the foreground information to obtain a support set prototype; and calculating a classification probability according to the support set prototype and the prior probability auxiliary information feature map to obtain a segmentation result. The method reduces over-segmentation and under-segmentation phenomena of large target organs, improves recognition of small target organs, can be used for multi-organ segmentation of medical images, and assists doctors in diagnosing diseases.

Description

technical field [0001] The invention belongs to the technical field of image processing, and mainly relates to a small-sample self-adaptive multi-organ segmentation method, which can be used for multi-target medical image processing of a small amount of label data in a complex environment. Background technique [0002] Multi-organ image segmentation is a very important research field in medical image processing, and is widely used in clinical auxiliary diagnosis and automated medical image analysis. In recent years, with the rapid development of deep learning technology, a large number of effective supervised learning-based segmentation networks have been applied to multi-organ segmentation tasks. At the same time, they have achieved labeling effects close to human experts on many multi-organ segmentation datasets. It greatly reduces the workload of clinicians and helps doctors analyze medical images efficiently. [0003] The current medical image segmentation model perform...

AI Technical Summary

Problems solved by technology

However, in the application of medical images, due to its characteristics different from natural images, the difference of single-channel pixel values ​​in different organs is smaller, resulting in less shape and size information extracted by the network, and large differences in the shape and size of target organs. and diversity, it also makes it difficult for the supervision information of the support set to fully guide the segmentation of the query set, so that the direct application of small sample learning cannot get good results

Images