Effectiveness of color correction on the quantitative analysis of histopathological images acquired by different whole-slide scanners

Maulana Abdul Aziz, Tomoya Nakamura, Masahiro Yamaguchi, Tomoharu Kiyuna, Yoshiko Yamashita, Tokiya Abe, Akinori Hashiguchi, Michiie Sakamoto

Abstract: Advances in whole-slide scanning have led to research on digital pathology, including monitor-based diagnosis, feature quantification, and computer-aided diagnosis. The color variations due to the staining process and scanning device are a serious issue that should be solved in whole-slide imaging applications, and methods for color correction have been studied. Nevertheless, the effectiveness of color-correction methods in the quantitative analysis of histopathological images acquired by different whole-slide scanners (WSSs) has not been confirmed. In this work, several liver tissue samples were scanned by different WSSs, and a color-correction method for whole-slide images was applied to the digitized tissue specimens. A set of histological features were extracted from the histopathological images, both without and with color correction, to analyze the effectiveness of the color-correction method in feature quantification and cancer identification.