In The American journal of pathology ; h5-index 54.0
Whole slide imaging (WSI) is becoming a routine procedure in clinical diagnosis. Advanced image analysis techniques have been developed to assist pathologists in disease diagnosis, staging, subtype classification and risk stratification. Recently, deep learning algorithms have achieved state-of-the-art performances in various imaging analysis tasks, including tumor region segmentation, nuclei detection, and disease classification. However, these algorithms are not widely applied in clinical applications, because their performances often degrade as a result of image quality issues, which are commonly seen in real world pathology imaging data, such as low resolution, blurring regions, and staining variation. To resolve these challenges, we developed Restore-GAN, a deep-learning model to improve the imaging qualities by restoring blurred regions, enhancing low resolution, and normalizing staining colors. Our results demonstrate that Restore-GAN can significantly improve image quality, which leads to improved model robustness and performance for existing deep learning algorithms in pathology image analysis. We envision that Restore-GAN can be used to facilitate the applications of deep learning models in digital pathology analyses.
Rong Ruichen, Wang Shidan, Zhang Xinyi, Wen Zhuoyu, Cheng Xian, Jia Liwei, Yang Donghan M, Xie Yang, Zhan Xiaowei, Xiao Guanghua
2023-Jan-17
Digital pathology, Generative adversarial network, color normalization, image enhancing, image restoration
