Research on Image Classification of Pathology Based on Deep Learning
Runcheng Wuzhao, Gao Gong, Shi Cao, Chaomin Chen, Wei Min Wang · Journal of intelligence and knowledge engineering. · 2024
Histopathological image classification based on in-depth study is one of the significant methods for diagnosing diseases. Pathology image information is complex, diverse, and has a large number of features, which leads to difficult classification and low diagnostic accuracy. Building computer-aided diagnostic tools using image processing and artificial intelligence techniques can dramatically improve pathologists’ productivity and reduce error rates. The classification methods for histopathological images mainly included the traditional machine learning methods based on manual feature extraction and deep learning-based pathology image classification methods. The classification accuracy of principal component analysis (PCA), random forest (RF), and support vector machine (SVM) in the traditional machine learning methods were 99.05%, 88%, and 80-85%, respectively. In the deep learning method, the classification accuracy of the convolutional neural network GoogleNet, AlexNet, and deep residual network ResNet is 91%, 99.74%, and 97.4%, respectively. Compared with the shortcomings of traditional classification methods of pathological images, the histopathology image classification methods of deep learning have obvious advantages, but there is also over-dependence on data labeling, feature extraction, and algorithm models, which need to be improved and improved in the future.