Melanoma Decision Support Using Lighting-Corrected Intuitive Feature Models

Abstract

Skin lesion analysis using standard camera images has received limited attention from the scientific community due to its technical complexity and scarcity of data. The images are privy to lighting variations caused by uneven source lighting, and unconstrained differences in resolution, scale, and equipment. In this chapter, we propose a framework that performs illumination correction and feature extraction on photographs of skin lesions acquired using standard consumer-grade cameras. We apply a multi-stage illumination correction algorithm and define a set of high-level intuitive features (HLIF) that quantifies the level of asymmetry and border irregularity about a lesion. This lighting-corrected intuitive feature model framework can be used to classify skin lesion diagnoses with high accuracy. The framework accurately corrects the illumination variations and achieves high and precise sensitivity (95 % confidence interval (CI), 73.1-73.5 %) and specificity (95 % CI, 72.0-72.4 %) using a linear support vector machine classifier with cross-validation trials. It exhibits higher test-retest reliability than the much larger state-of-the-art low-level feature set (95 % CI,. Combining our framework with these low-level features attains sensitivity (95 % CI, 83.3-84.8 %) and specificity (95 % CI, 79.7-80.1 %), which is more accurate and reliable than classification using the low-level feature set.

Performance Analysis of Low-Level and High-Level Intuitive Features for Melanoma Detection

“…The performance is better in terms of sensitivity and comparable results in terms of specificity. Sensitivity in the [31] is 90%, while in our proposed method the sensitivity found is 95.8%. Specificity in previous work is 96%, while, in our proposed method, it is 91%.…”

“…The results are derived on publicly available datasets of DermIS and DermQuest with 206 images. The dataset used in literature as stated in Section 2 in [31] is identical to the one used in our research paper, i.e., DermIS and DermQuest with the same number of images i.e., 206. In [31], they found 84.04%, 79.91% and 81.26% of sensitivity, specificity and accuracy, respectively.…”

“…Amelard et al [31] applied a multi-stage illumination correction algorithm on dermIS and dermQuest datasets comprised of 206 images and analyzed the low-level and high-level features. They have taken nine high-level intuitive features and 52 Cavalcanti and Scharcanski's low-level features.…”

“…The number of images used in [31] is 200 images, while in our paper we used 206 images. The number of images used in both papers is the same, but the datasets are different.…”

“…The bottom line is that, in both of the papers, ABCD features are incorporated. In [31], a subset of ABCD features is combined with GLCM parameters along with the CFS algorithm to select features. In contrast, we used a complete set of ABCD features with GLCM parameters, although the performance is not significant.…”