Fully Convolutional Genetic Neural Network Method for Image Segmentation of Infant Brain Medical Records

Abstract

The invention discloses a fully convolutional genetic neural network method for image segmentation of infant brain medical records, which belongs to the field of medical image information intelligent processing. The network weight length L is used to initialize the genetic coding of the parameters; then m individuals are randomly divided into the genetic primordial subpopulation Pop and the twin subpopulation Pop' is derived, and the subpopulations determine their respective exchange probability p in the disjoint interval c and the mutation probability p m , use the genetic operator to find the optimal initial weight fa; secondly, use fa as the forward propagation calculation parameter, and perform a weighted Q operation on the feature address featuremap; finally, the predicted image of the infant brain medical record and the standard segmentation map are crossed pixel by pixel The entropy loss is calculated to update the weights in reverse, and finally the optimal weights of the image segmentation network model of infant brain medical records are obtained. The method can improve the image segmentation efficiency of infant encephalopathy records, and is of great significance to the early correct diagnosis of infant encephalopathy and the rehabilitation of infant encephalopathy.

Description

technical field [0001] The invention relates to the field of intelligent processing of medical image information, in particular to a fully convolutional genetic neural network method for image segmentation of infant brain medical records. Background technique [0002] Infancy is a critical period for brain development. It not only develops rapidly, but also has strong plasticity. The probability of suffering from various encephalopathy is also much higher than that of adults or other growth periods. In recent years, the survival rate of high-risk infants and very low-birth-weight infants has increased significantly, and the incidence of brain diseases in infants and young children has shown an upward trend, making early diagnosis more difficult. Therefore, actively exploring computer intelligence-assisted early diagnosis methods for infantile encephalopathy is of great significance to the diagnosis and evaluation of infantile encephalopathy. [0003] In recent years, with t...

AI Technical Summary

Problems solved by technology

However, due to the characteristics of low resolution and non-uniform gray scale of the images of infant brain medical records, when the traditional full convolutional neural network method is used to segment the images of infant brain medical records, gradient descent is prone to occur, and the algorithm is easy to fall into local areas. Notable problems such as the optimal solution and image feature information cannot be effectively preserved, which lead to problems such as time-consuming, difficult training, and low precision in the image segmentation of infant brain medical records by these methods, so that high-accuracy infant brain medical record segmentation cannot be obtained. image

Images