Intrinsic Melanin and Hemoglobin Colour Components for Skin Lesion Malignancy Detection

Abstract

Abstract. In this paper we propose a new log-chromaticity 2-D colour space, an extension of previous approaches, which succeeds in removing confounding factors from dermoscopic images: (i) the effects of the particular camera characteristics for the camera system used in forming RGB images; (ii) the colour of the light used in the dermoscope; (iii) shading induced by imaging non-flat skin surfaces; (iv) and light intensity, removing the effect of light-intensity falloff toward the edges of the dermoscopic image. In the context of a blind source separation of the underlying colour, we arrive at intrinsic melanin and hemoglobin images, whose properties are then used in supervised learning to achieve excellent malignant vs. benign skin lesion classification. In addition, we propose using the geometric-mean of colour for skin lesion segmentation based on simple greylevel thresholding, with results outperforming the state of the art.

Melanin and Hemoglobin Identification for Skin Disease Analysis

“…Based on this multilayered model, previous methods [2,3] use Lambert-Beer law for skin radiance provided that there is little effect of specular reflection. This diffuse reflectance model is effective in dealing with dermoscopy images under polarized lighting, but it is hard to handle conventional RGB skin images captured in uncontrolled imaging setting.…”

“…Tsumura et al [2] applied independent component analysis (ICA) on RGB skin image to separate average concentrations of melanin and hemoglobin. Madooei et al [3] extended their work by introducing extra factors to spectral reflection of skin model to reduce external effects such as shading and sensor characteristics. These methods show effectiveness of describing skin coloration for melanoma diagnosis.…”

“…More importantly, all of the above methods either do not have/specify an illumination and shading removal step [2,4,5], or cancel out the effects of illumination and shading by directly dividing two spectral responses, provided that there is the same amount of external artifacts embedded in different spectral images [3]. But illumination spectrum varies with wavelength.…”

Melanoma Decision Support Using Lighting-Corrected Intuitive Feature Models

“…D'Alessandro et al used multispectral images obtained using a nevoscope and used a genetic algorithm to estimate melanin and blood volume [20]. Madooei et al used blind-source separation techniques using a proposed corrected log-chromaticity 2-D colour space to obtain melanin and hemoglobin information [38].…”

Automatic Detection of Blue-White Veil by Discrete Colour Matching in Dermoscopy Images

“…This indicates that their method can benefit from the choice of colour representation. To investigate this, we have reproduced their training experiment where each pixel has been represented by its corresponding coordinates in RGB, nRGB, L*a*b*, YIQ, HSV, and MHG [11] colour spaces. These colour models were chosen since each represents a different family of colour spaces.…”