An ICA-Based Method for the Segmentation of Pigmented Skin Lesions in Macroscopic Images

PABLO CAVALCANTI; JACOB SCHARCANSKI; LEANDRO EZEQUIEL DI PERSIA; DIEGO HUMBERTO MILONE

Boston

Congreso; 33rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC ’11).; 2011

IEEE

Segmentation is an important step in computeraideddiagnostic systems for pigmented skin lesions, since thata good definition of the lesion area and its boundary at theimage is very important to distinguish benign from malignantcases. In this paper a new skin lesion segmentation method isproposed. This method uses Independent Component Analysisto locate skin lesions in the image, and this location informationis further refined by a Level-set segmentation method. Ourmethod was evaluated in 141 images and achieved an averagesegmentation error of 16.55%, lower than the results forcomparable state-of-the-art methods proposed in literature.