A Histopathological Image Feature Representation Method based on Deep Learning

Automated annotation and grading for histopathological image plays an important role in CAD systems. It provides valuable information and support for medical diagnosis. Currently, computer-aid analysis of histopathological images mainly relies on some well-designed digital features, which requires abundant human efforts and experiences in problem domain. Learning a good feature representation from data can have positive effects on constructing the target model. We propose a novel method for histopathological image feature representation based on deep learning. The method extracts high level representation of raw pixels of a local region through a network model with several hidden layers, which can learn potential features automatically. The proposed method is evaluated on a real data set from a large local hospital with comparison to two current state-of-the-art methods. The result is promising indicating that it achieves significant improvement of the model performance. Moreover, our study suggests that features learned through deep models can achieve better performance than human designed features.