Optical Transfer Diagnosis of Pigmented Lesions

Abstract

This technology continues to be a promising adjunct to clinical skin cancer screening.

Optical detection and monitoring of pigmented skin lesions

“…Preliminary results based on analyses of independent measurements performed on a limited set of suspicious lesions [15,16] indicated that a linear binary classifier would suffice to discriminate between independent measurements on malignant (class 1) lesions and benign (class 2) lesions. But as the set of suspicious lesions was enlarged with the same type of lesions, a linear binary classifier was found not to perform satisfactorily, motivating the development and use of clustering, which is used to obtain discrimination between class 1 and class 2 lesions through the identification of one set of class 1 clusters and another set of class 2 clusters, each cluster comprising a certain number of independent measurements.…”

“…In two previous studies [15,16], the application of optical transfer diagnosis (OTD) for noninvasive detection of skin cancer was discussed, the objective being to evaluate the potential of this technology for differentiation of benign from malignant pigmented melanocytic lesions. After scanning, lesions were biopsied for histopathologic examination, each by two separate dermatopathologists.…”

“…After scanning, lesions were biopsied for histopathologic examination, each by two separate dermatopathologists. The second study [16] included 94 patients with a total of 118 pigmented lesions suggestive of melanoma, out of which 11 were identified as melanoma or atypical melanocytic hyperplasia consistent with melanoma. For identification of melanomas, OTD had a sensitivity of 100% and a specificity of 90%.…”

“…The results in [15,16] were based on using a linear binary classifier to differentiate between malignant and benign lesions in a population of suspicious lesions. As the set of measurements on suspicious lesions was enlarged with similar types of lesions beyond the 118 lesions studied in [16], a linear binary classifier was found not to perform satisfactorily, probably due to overtraining on the limited initial data set, providing overoptimistic results that could not be sustained as the data set was enlarged. The objective of the new classification scheme discussed in this paper, which involves clustering and construction of nonlinear classifiers, is to discriminate between two classes of tissue, e.g.…”

“…The objective of this paper is to describe new features of the OTD classification scheme compared to those presented in [15,16], and present results that demonstrate its potential.…”