Incorporating Colour Information for Computer-Aided Diagnosis of Melanoma from Dermoscopy Images

Cutaneous Melanoma is the most life-threatening form of skin cancer. Although advanced melanoma is often considered as incurable, if detected and excised early, the prognosis is promising. Today, clinician use Computer Vision in an increasing number of applications to aid early detection of melanoma through dermatological image analysis (dermoscopy images, in particular). Colour assessment is essential for the clinical diagnosis of skin cancers. Due to this diagnostic importance, many studies have either focused on or employed colour features as a constituent part of their skin lesion analysis systems. These studies range from using low-level colour features, such as simple statistical measures of colours occurring in the lesion, to availing themselves of high-level semantic features such as the presence of blue-white veil, globules or colour variegation in the lesion. This thesis provides a detailed exposition of my recent contributions in this research direction. In particular, it describes two novel approaches for utilizing colour both as low-level and high- level image feature. The first contribution describes a technique that employs the stochastic Latent Topic Models framework to allow quantification of melanin and hemoglobin content in dermoscopy images. Such information bears useful implications for the analysis of skin hyper-pigmentation, and for classification of skin diseases. The second contribution is a novel approach to identify one of the most significant dermoscopic criteria in the diagnosis of Cutaneous Melanoma: the Blue-whitish structure. This is achieved in a Multiple Instance Learning framework with only image-level labels of whether the feature is present or not. As the output, we predict the image label and also localize the feature in the image. Experiments are conducted on a challenging dataset with results outperforming state-of-the-art. Moreover, the thesis explores using physic-based photometric models to enhance dermoscopy image analysis. In particular, it proposes methods for colour-to-greyscale conversion, shading removal, and glare attenuation. The studies reported in this thesis provide an improvement on the scope of modelling for computerized image analysis of skin lesions