Portal de Periódicos da CAPES

Computer-aided classification of melanocytic lesions using dermoscopic images

2015; Elsevier BV; Volume: 73; Issue: 5 Linguagem: Inglês

10.1016/j.jaad.2015.07.028

1097-6787

Laura K. Ferris, Jan Harkes, Ben Gilbert, Daniel G. Winger, Kseniya Golubets, Oleg E. Akilov, Mahadev Satyanarayanan,

AI in cancer detection

Background Computer-assisted diagnosis of dermoscopic images of skin lesions has the potential to improve melanoma early detection. Objective We sought to evaluate the performance of a novel classifier that uses decision forest classification of dermoscopic images to generate a lesion severity score. Methods Severity scores were calculated for 173 dermoscopic images of skin lesions with known histologic diagnosis (39 melanomas, 14 nonmelanoma skin cancers, and 120 benign lesions). A threshold score was used to measure classifier sensitivity and specificity. A reader study was conducted to compare the sensitivity and specificity of the classifier with those of 30 dermatology clinicians. Results The classifier sensitivity for melanoma was 97.4%; specificity was 44.2% in a test set of images. In the reader study, the classifier's sensitivity to melanoma was higher ( P < .001) and specificity was lower ( P < .001) than that of clinicians. Limitations This is a retrospective study using existing images primarily chosen for biopsy by a dermatologist. The size of the test set is small. Conclusions Our classifier may aid clinicians in deciding if a skin lesion should be biopsied and can easily be incorporated into a portable tool (that uses no proprietary equipment) that could aid clinicians in noninvasively evaluating cutaneous lesions.