Analysis of Three Point Checklist and ABCD Methods for the Feature Extraction of Dermoscopic Images to Detect Melanoma

A man's danger of causing malignancy depends upon numerous components such as age, hereditary qualities, risk to chance variables. There has been a progressive increase in the occurrence of skin cancers over the last few decades. So, a Computer Aided Diagnostic (CAD) system is highly in demand to assist doctors to reduce their physical effort for the melanoma detection. This work aims to develop efficient feature extraction algorithms for classification of melanoma and non-melanoma cases to develop a CAD system. The algorithms are verified via images available in the PH 2 dataset. The ABCD and 3 point checklist features are extracted from the preprocessed images. These features are analysed and compared using various classification algorithms such as Sequential Minimal Optimisation (SMO), Logistic regression, Random Forest, J48 decision tree available in WEKA software. In the analysis, it is found that 3 - Point checklist has a better accuracy and precision compared to ABCD method for various classifiers tested in Weka 3.8.2 software.