Comparative Study on Boundary Structural Irregularity Using Local FD and Curvature Analysis for Melanoma Detection

Feature extracted from structural irregularity for skin lesion boundaries has a great significance in computer-aided diagnosis for melanomas. Based on previous work using local fractal dimension (local FD) for contour irregularity descriptions, the novelty of this paper focuses on: (1) Multi-scaled curvature analysis is used to acquire features of boundary irregularity (2) Feature differences from adjacent scales are investigated to enhance the discrimination power between malignant skin tumors and benign. The comparative experiments of both local FD and curvature analysis show that the features difference of adjacent scales has characteristics of larger inter-class Hausdorff distance with the scale increase, while local FD performs better than curvature analysis with larger Hausdorff distance value. Results show that these two approaches not only have strong classification capacity of structure irregularity, but also helpful to remove texture irregularity for achieving robust classification.