Romero-Lopez A, Burdick J, Giró-i-Nieto X, Marques O. The Impact of Segmentation on the Accuracy and Sensitivity of a Melanoma Classifier based on Skin Lesion Images. In Annual Meeting of the Society of Imaging Informatics in Medicine (SIIM). Pittsburgh, PA, USA: Society of Imaging Informatics for Medicine; 2017.  (433.28 KB)


The accuracy and sensitivity of a Deep Learning based approach for a 2-class classifier for early melanoma detection based on skin lesion dermoscopic images increases when the classifier is trained with segmented inputs (i.e., images containing only the lesions as binary masks, without the surrounding context) instead of entire images.

[SIIM 2017 Annual Meeting website]

[SIIM 2017 Session where our work is presented]