Melanoma Detection Using CBR Approach Within a Possibilistic Framework

Solar UltraViolet Radiation (UVR) exposure is clearly associated with increased mortality from melanoma cancer cases in recent years, with this disease exhibiting the highest mortality rates among dermatological cancers. However, recent research indicates that early diagnosis significantly improves life expectancy. While dermoscopic images require specialized devices for acquisition, utilizing optical skin images captured with a standard camera emerges as a cost-effective and viable method for early melanoma detection, demonstrating satisfactory detection rates. This study introduces an innovative approach that integrates case-based reasoning into a possibility theory framework, employing optical skin lesion images to aid experts in the early detection of melanoma. Experimental validation on optical lesion image datasets demonstrates the suitability of our approach for lesion severity classification, achieving a specificity of 100% and an accuracy of 95%, surpassing the performance of recent methods in melanoma diagnosis.