BACKGROUND: Differential diagnosis of melanoma from melanocytic nevi is often not straightforward. Thus, a growing interest has developed in the last decade in the automated analysis of digitized images obtained by epiluminescence microscopy techniques to assist clinicians in differentiating early melanoma from benign skin lesions. PURPOSE: The aim of this study was to evaluate diagnostic accuracy provided by different statistical classifiers on a large set of pigmented skin lesions grabbed by four digital analyzers located in two different dermatological units. EXPERIMENTAL DESIGN: Images of 391 melanomas and 449 melanocytic nevi were included in the study. A linear classifier was built by using the method of receiver operating characteristic curves to identify a threshold value for a fixed sensitivity of 95%. A K-nearest-neighbor classifier, a nonparametric method of pattern recognition, was constructed using all available image features and trained for a sensitivity of 98% on a large exemplar set of lesions. RESULTS: On independent test sets of lesions, the linear classifier and the K-nearest-neighbor classifier produced a mean sensitivity of 95% and 98% and a mean specificity of 78% and of 79%, respectively. CONCLUSIONS: In conclusion, our study suggests that computer-aided differentiation of melanoma from benign pigmented lesions obtained with DB-Mips is feasible and, above all, reliable. In fact, the same instrumentations used in different units provided similar diagnostic accuracy. Whether this would improve early diagnosis of melanoma and/or reducing unnecessary surgery needs to be demonstrated by a randomized clinical trial.
BACKGROUND: Differential diagnosis of melanoma from melanocytic nevi is often not straightforward. Thus, a growing interest has developed in the last decade in the automated analysis of digitized images obtained by epiluminescence microscopy techniques to assist clinicians in differentiating early melanoma from benign skin lesions. PURPOSE: The aim of this study was to evaluate diagnostic accuracy provided by different statistical classifiers on a large set of pigmented skin lesions grabbed by four digital analyzers located in two different dermatological units. EXPERIMENTAL DESIGN: Images of 391 melanomas and 449 melanocytic nevi were included in the study. A linear classifier was built by using the method of receiver operating characteristic curves to identify a threshold value for a fixed sensitivity of 95%. A K-nearest-neighbor classifier, a nonparametric method of pattern recognition, was constructed using all available image features and trained for a sensitivity of 98% on a large exemplar set of lesions. RESULTS: On independent test sets of lesions, the linear classifier and the K-nearest-neighbor classifier produced a mean sensitivity of 95% and 98% and a mean specificity of 78% and of 79%, respectively. CONCLUSIONS: In conclusion, our study suggests that computer-aided differentiation of melanoma from benign pigmented lesions obtained with DB-Mips is feasible and, above all, reliable. In fact, the same instrumentations used in different units provided similar diagnostic accuracy. Whether this would improve early diagnosis of melanoma and/or reducing unnecessary surgery needs to be demonstrated by a randomized clinical trial.
Authors: Nikita V Orlov; Ashani T Weeraratna; Stephen M Hewitt; Christopher E Coletta; John D Delaney; D Mark Eckley; Lior Shamir; Ilya G Goldberg Journal: Cytometry A Date: 2012-03-29 Impact factor: 4.355
Authors: Andre Esteva; Brett Kuprel; Roberto A Novoa; Justin Ko; Susan M Swetter; Helen M Blau; Sebastian Thrun Journal: Nature Date: 2017-01-25 Impact factor: 49.962
Authors: Lavinia Ferrante di Ruffano; Yemisi Takwoingi; Jacqueline Dinnes; Naomi Chuchu; Susan E Bayliss; Clare Davenport; Rubeta N Matin; Kathie Godfrey; Colette O'Sullivan; Abha Gulati; Sue Ann Chan; Alana Durack; Susan O'Connell; Matthew D Gardiner; Jeffrey Bamber; Jonathan J Deeks; Hywel C Williams Journal: Cochrane Database Syst Rev Date: 2018-12-04