Global color moments and RGB/HSV histograms alone support binary benign-malignant classification at up to 89% accuracy with classical ML classifiers, substantially above random baselines.
The impact of site-specific digital histology signatures on deep learning model accuracy and bias.Nature communications, 12(1):4423, 2021
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Beyond Morphology: Quantifying the Diagnostic Power of Color Features in Cancer Classification
Global color moments and RGB/HSV histograms alone support binary benign-malignant classification at up to 89% accuracy with classical ML classifiers, substantially above random baselines.