Thyroid nuclear medicine image diagnosis method based on deep learning

Abstract

The invention discloses a thyroid nuclear medicine image diagnosis method based on deep learning, and belongs to the technical field of information. The problem that thyroid nuclear medicine image diagnosis is limited to a certain extent and the diagnosis efficiency is low in the prior art is solved. The method comprises the following steps: determining an ROI area by using a three-channel threshold detection method, and cutting pictures; performing scaling, rotation and Gaussian noise addition processing on the focus image set to form an extended sample data set; establishing a convolutionalneural network for image diagnosis, and training the network by using the training set to obtain a trained novel diagnosis model; performing disease diagnosis on the test set by using the trained diagnosis model to obtain a thyroid disease diagnosis result of each medical image. According to the method, the trained model has higher robustness, and the popularization possibility of the model in hospitals is enhanced.

Description

technical field [0001] The present application belongs to an image diagnosis method, belongs to the field of information technology, and specifically relates to an image diagnosis method of thyroid nuclear medicine based on deep learning. Background technique [0002] The thyroid is a hormone-producing gland located in the lower front of the neck that controls the rate at which energy is used, makes protein, regulates the body's sensitivity to other hormones, regulates metabolism, growth rate, and regulates other bodily processes. In recent years, the incidence of thyroid cancer in various countries has gradually increased, and the global incidence is increasing at an annual rate of 6.2%. In China, the incidence of thyroid cancer is showing an obvious upward trend. The thyroid is the largest endocrine gland in the human body, and its disease detection rate is increasing day by day, and this trend is more obvious with the increase of age. At present, SPECT medical imaging is...

AI Technical Summary

Problems solved by technology

In the process of thyroid disease diagnosis, the essential texture details of the problem object determine the configuration of the network. The robustness of the deep learning model has certain limitations, and the diagnostic efficiency is low, so it is not suitable for promotion in hospitals.

Images