Stain independent nuclei segmentation of cytopathology images: A case study in pleural effusion

Digital pathology (DP) is a technology that makes it possible to apply image analysis techniques to cytopathology images for cancer diagnosis. Color variation is a common problem in cytopathology as a result of a number of factors, including variable chemical coloring of different stain manufacturers, and inconsistencies in the staining procedure. In this study, a new stain normalization method is proposed using the residual learning approach. The performance of the proposed method was measured by applying the problem of nuclei segmentation in pleural effusion (PE) cytopathology images. The proposed method has significantly improved the performance of image analysis techniques that are sensitive to color variations.