改进残差网络的医学X射线影像分类与加密传输系统

Medical X-ray Image Classification and Encrypted Transmission System Based on Improved Residual Network

随着X射线影像在医疗诊断领域的快速发展,在大量胸腔X射线影像产出的情况下,医生根据经验进行人为判断分析的方式已不能满足诊断效率与准确率的需求,高效率、高准确率处理批量X射线影像分类的问题亟待解决;通过改进残差网络对胸腔X射线影像进行分类,并设计一种加密传输系统,可有效解决上述问题;利用对X射线影像进行基于马尔可夫随机场的图像增强,再采用深层信息挖掘能力较强的ResNet50作为主干网络,增加自注意力机制并采用CELU激活函数优化;经Kaggle整合数据集实验测试结果表明,在保证分类准确性的前提下,分类的召回率从0.432提升到0.652;同时,系统采用基于Logistic混沌序列的图像加密算法,保证了远程医疗诊断的私密性,满足实际远程医疗场景的应用需求。

With the rapid development of X-ray imaging in the field of medical diagnosis,traditional manual judgment and analysis methods based on the doctor's experience cannot meet the efficiency requirements for diagnosing a large number of chest X-ray images.it is urgent to efficiently and accurately solve the classification of batch X-ray images.The residual network is introduced to classify the batch X-ray images,and an encrypted transmission system is designed to effectively achieve the above problem.The X-ray images are used to enhance the images based on the Markov random field,and then the ResNet50 with strong deep information mining ability is used as the backbone network to add the self-attention mechanism,and the continuously differentiable exponential linear units(CELU) activation function is used to optimize.The experiment is implemented on the Kaggle integrated dataset,the results show that the recall rate of classification is increased from 0.432 to 0.652 while ensuring the classification accuracy.Additionally,the image encryption algorithm based on logistic chaotic sequences is used to ensure the privacy of remote medical diagnosis,meeting the application requirements of actual remote medical scenarios.