基于CNN和KNN的镰状细胞图像分类模型

Sickle Cell Image Classification Algorithm based on CNN and KNN

构建基于卷积神经网络(CNN)和K-最近邻(KNN)的混合智能分类模型——CNN-KNN模型,旨在自动识别镰状细胞病(SCD),提高分类精度.通过预处理技术增强红细胞图像特征,采用模糊C-均值聚类与自适应阈值方法分割细胞;设计包含24层的深度卷积神经网络(CNN)模型,通过训练得到卷积层和全连接层参数;将KNN分类器与CNN模型结合,构建CNN-KNN模型.实验结果表明,CNN-KNN模型表现出极佳的性能.

A hybrid intelligent classification model based on convolutional neural network(CNN)and K-nearest neighbor(KNN),CNN-KNN model,is constructed with the aim of automatically identifying sickle cell disease(SCD)and improving the classification accuracy.First,the erythrocyte image features are enhanced by preprocessing techniques,and the cells are segmented using fuzzy C-mean clustering with adaptive thresholding method.Second,a deep convolutional neural network(CNN)model containing 24 layers is designed,and the parameters of convolutional and fully connected layers are obtained through training.Finally,the KNN classifier is combined with the CNN model to construct the CNN-KNN model.The experimental results show that the CNN-KNN model exhibits excellent performance.