Beyond Global Shannon Entropy: A Channel-Specific Approach to Quantify Polychromia in Melanoma

Area of Science:

• Dermatology
• Image Analysis
• Computational Pathology

Background:

• Polychromia is a key dermoscopic melanoma indicator, but its assessment is subjective.
• Shannon entropy offers objective color heterogeneity measurement, but global entropy may not capture chromatic complexity.
• Distinguishing melanoma from benign nevi requires accurate color analysis.

Purpose of the Study:

• To evaluate if global Shannon entropy quantifies polychromia in smartphone images.
• To determine if channel-specific entropy metrics better reflect chromatic heterogeneity.
• To assess the utility of these metrics for discriminating melanoma from nevi.

Main Methods:

• Smartphone images of melanoma, nevus, and perilesional skin were analyzed.
• Shannon entropy was computed for grayscale, RGB, and individual color channels (R, G, B).
• Metrics were normalized to perilesional skin to control for illumination and baseline heterogeneity.

Main Results:

• Global entropy from grayscale and RGB histograms showed similar values, reflecting luminance, not chromaticity.
• Channel-specific analysis revealed significant chromatic heterogeneity in melanoma vs. nevus (e.g., ΔG-B_residual = +12.31).
• A normalized Polychromia Index showed clear separation between melanoma (+6.84) and nevus (-1.38).

Conclusions:

• Global Shannon entropy is inadequate for quantifying polychromia in smartphone images.
• Channel-specific entropy and inter-channel metrics reliably differentiate chromatically heterogeneous lesions.
• This framework provides objective color assessment for teledermatology and automated melanoma detection.