Beyond Morphology: Quantifying the Diagnostic Power of Color Features in Cancer Classification

In three linesarXiv study demonstrates that color features alone (RGB/HSV histograms, statistical moments) achieve 89% accuracy in binary cancer/benign classification in histopathology, excluding morphological information. Authors propose these simple features as lightweight pre-screening tool before complex deep learning models.

Summary generated by Claude — human-verified