STN-based autism brain magnetic resonance image visualization method

Abstract

The invention discloses an STN-based autistic brain magnetic resonance image visualization method, which comprises the following steps: acquiring a brain nuclear magnetic resonance image of an autistic patient as a training sample, and training an STN model by using the training sample to obtain a transformed image; and training the transformed image according to the STN model, and transmitting thetransformed image to a convolutional neural network for training, so that visualization of the brain nuclear magnetic resonance image of the autistic patient is achieved. According to the STN model,the autism brain magnetic resonance image is processed on the basis of the convolutional neural network model, automatic recognition and detection are achieved, the visualization effect is good, the STN model is obviously improved compared with the convolutional neural network model, and medical researchers can be assisted in quantitative analysis and research conveniently.

Description

technical field [0001] The invention belongs to the technical field of nuclear magnetic resonance image disease visualization, and in particular relates to an STN-based visualization method for brain magnetic resonance images of autism. Background technique [0002] In recent years, the field of computer vision has undergone earth-shaking changes and has continued to move forward. The convolutional neural network model has achieved good results in classification, target detection, image segmentation, and action recognition. Convolutional neural networks define a very powerful class of models, but are still limited by the lack of invariance to input images due to the usually small transformation range of max pooling invariance to translation, scaling, rotation and more general deformations. [0003] In this case, the spatial transformation network (STN, Spatial Transformer Network) module can be included in the existing convolutional neural network architecture. Unlike the po...

AI Technical Summary

Problems solved by technology

Convolutional neural networks define a very powerful class of models, but are still limited by the lack of invariance to input images due to the usually small transformation range of max-pooling invariance to translation, scaling, rotation, and more general deformations.

Images