Multi-Channel Color Space Analysis for Nuclei Segmentation in Histopathology Images

Nuclei segmentation is essential in various medical analysis applications to assess targeted diseased tissue. In histopathology images, the unsupervised learning method widely used for segmenting cell nuclei is based on processing color histopathology images using standard k-means or fuzzy c-means (FCM) algorithms. Previous studies have explored the significant impact of different color spaces on segmentation results. However, under the background of unsupervised cell nuclei segmentation in histopathology images, a research gap exists concerning the integration of multiple color spaces and the utilization of diverse channel combinations within them. Our study aims to investigate the integration of suitable color spaces for enhancing cell nuclei segmentation results. Specifically, we select two color spaces that demonstrate superior segmentation outcomes. To assess the efficacy of nuclei segmentation, we use one of these color spaces as a reference and randomly combine channels from the other color spaces. Through the exploration of different channel combinations, we can leverage their potential synergistic effects to enhance the segmentation performance. The experimental results indicate that the superior nuclei segmentation result is achieved by combining the entire L*a*b color space with the Cb and Cr channels in the YCbCr color space. This finding emphasizes the advantages of integrating multiple color spaces to improve the segmentation outcomes.