Evaluación de técnicas de normalización de color para reducción de heterogeneidad de tejidos en imágenes histopatológicas de cáncer de mama

Color normalization in histopathological images is considered an initial step aimed at contributing to consistency in the computational analysis of hematoxylin and eosin (H&E)-stained tissues. In this work, three color normalization methods are evaluated in histopathological images of breast cancer (Macenko, Reinhard, and Vahadane) using quantitative metrics ORB, SSIM, and histogram correlation. In terms of ORB and SSIM, Vahadane stood out for better preserving morphological structures with consistently high values, which is crucial for computational analysis. Nevertheless, Vahadane presented variability in histogram correlation, in contrast to Macenko and Reinhard, who achieved higher and consistent correlations across all samples with average values of 0.9628 and 0.9385. This highlights that the choice of method depends on the balance between structural fidelity and chromatic distribution depending on the case.