Clinically Inspired Skin Lesion Classification through the Detection of Dermoscopic Criteria for Basal Cell Carcinoma

Abstract

Background and Objective. Skin cancer is the most common cancer worldwide. One of the most common non-melanoma tumors is basal cell carcinoma (BCC), which accounts for 75% of all skin cancers. There are many benign lesions that can be confused with these types of cancers, leading to unnecessary biopsies. In this paper, a new method to identify the different BCC dermoscopic patterns present in a skin lesion is presented. In addition, this information is applied to classify skin lesions into BCC and non-BCC. Methods. The proposed method combines the information provided by the original dermoscopic image, introduced in a convolutional neural network (CNN), with deep and handcrafted features extracted from color and texture analysis of the image. This color analysis is performed by transforming the image into a uniform color space and into a color appearance model. To demonstrate the validity of the method, a comparison between the classification obtained employing exclusively a CNN with the original image as input and the classification with additional color and texture features is presented. Furthermore, an exhaustive comparison of classification employing different color and texture measures derived from different color spaces is presented. Results. Results show that the classifier with additional color and texture features outperforms a CNN whose input is only the original image. Another important achievement is that a new color cooccurrence matrix, proposed in this paper, improves the results obtained with other texture measures. Finally, sensitivity of 0.99, specificity of 0.94 and accuracy of 0.97 are achieved when lesions are classified into BCC or non-BCC. Conclusions. To the best of our knowledge, this is the first time that a methodology to detect all the possible patterns that can be present in a BCC lesion is proposed. This detection leads to a clinically explainable classification into BCC and non-BCC lesions. In this sense, the classification of the proposed tool is based on the detection of the dermoscopic features that dermatologists employ for their diagnosis.

Keywords: basal cell carcinoma; clinically inspired classification; color appearance models; color cooccurrence matrix; deep learning; dermatology.

Figure 1

ISIC BCC image samples (Challenge 2018).

Figure 2

Top row, from left to right: telangiectasia, multiple B/G globules, ulceration and pigment network; bottom row: spoke-wheel, blue-gray ovoids and maple leaf.

Figure 3

Representative colors of each BCC dermoscopic pattern in RGB color space. (a) Pigment network, (b) ulceration, (c) blue-gray ovoids, (d) multiple B/G globules, (e) maple leaf, (f) spoke-wheel, (g) telangiectasia, (h) the 20 final color centroids selected.

Figure 4

Color quantization in different color spaces. (a) Original image. (b) Image quantized in RGB color space (20 colors). (c) Image quantized in L*a*b* color space (18 colors). (d) Image quantized in CIECAM16-UCS (18 colors).

Figure 5

Dual classification. Original RGB and color-quantized images are used as inputs to the VGG16 CNNs, and concatenated to enter in a MLP classifier.

Figure 6

Triple classification. Original RGB, color-quantized images and texture descriptors are used as inputs to the MLP classifier.

Figure 7

Flow diagram followed by the participants in the four tests analyzed in this work.

Figure 8

Confusion matrix of the classification into BCC versus non-BCC.

Figure 9

ROC curve when classifying BCC versus non-BCC.