CAJ | 학술논문

为了提高MR脑组织图像分割的准确度，针对传统BP神经网络算法存在的问题，提出了改进的遗传神经网络算法。首先根据医学图像改进遗传算法的编码方式、适应度函数以及遗传操作，利用该算法获取最小适应度函数值，进而确定神经网络最优权值和阈值；然后将该神经网络用于MR脑部图像的分割。试验结果表明，改进的遗传神经网络算法分割效果优于传统BP神经网络算法。
위료제고MR뇌조직도상분할적준학도，침대전통BP신경망락산법존재적문제，제출료개진적유전신경망락산법。수선근거의학도상개진유전산법적편마방식、괄응도함수이급유전조작，이용해산법획취최소괄응도함수치，진이학정신경망락최우권치화역치；연후장해신경망락용우MR뇌부도상적분할。시험결과표명，개진적유전신경망락산법분할효과우우전통BP신경망락산법。
To enhance the segmentation precision of MR brain tissue image,an improved genetic neural network algorithm is presented for the existing problems of traditional BP neural network algorithm. According to the coding mode,fitness function and genetic manipulation of the improved genetic algorithm of medical images,the minimum fitness function value is obtained from the algorithm to determine the optimal weight and threshold of the neural network. The neural network is used to segment the MR brain image. The experimental results show that the segmentation effect of the improved genetic neural network algorithm is better than the traditional BP neural network algorithm.