Fuzzy logic techniques for blotch feature evaluation in dermoscopy images

Area of Science:

• Dermatology and Medical Image Analysis
• Computational Intelligence in Healthcare

Background:

• Blotches, or structureless areas, are key indicators for distinguishing malignant melanoma from benign skin lesions in dermoscopy images.
• Accurate identification of these features is crucial for early cancer detection and effective treatment planning.

Purpose of the Study:

• To investigate fuzzy logic techniques for the automatic detection of blotch features in dermoscopy images.
• To enhance the discrimination accuracy of malignant melanoma using these automated blotch features.

Main Methods:

• Utilized four fuzzy sets representing blotch size and relative/absolute colors to extract blotchy areas from dermoscopy images.
• Computed five established and four novel blotch features from the extracted areas.
• Employed a neural network classifier and optimized results across various alpha-cuts, comparing fuzzy versus crisp blotch features.

Main Results:

• Fuzzy logic techniques, specifically analyzing three-plane relative color and blotch size, yielded the highest diagnostic accuracy.
• The optimized fuzzy logic approach achieved a diagnostic accuracy of 81.2% for malignant melanoma discrimination.

Conclusions:

• Fuzzy logic-based feature extraction significantly improves the accuracy of differentiating malignant melanoma from benign lesions.
• The developed method offers a promising automated approach for enhancing diagnostic capabilities in dermatological image analysis.