Skin-Tumour Classification with Functional Infrared Imaging

The number of skin-cancer cases has dramatically increased during the last decades. Generally, an early diagnosis is the key for successful treatment. However, a first diagnosis is always done with Stolz's traditional (subjective) ABCD rule of dermatoscopy based on the four main criteria or lesion parameters: Asymmetry, Border, Colour and Diameter. Unfortunately, a sound decision if a skin tumor is malignant or benign is based on an invasive histological examination only. To avoid redundant excise of tissue and to detect malignant skin lesions in an early stage, a new method is described here. Functional IR imaging is proposed for differential diagnosis of skin tumour types. The principle follows the paradigm that the temperature signature of the thermoregulation process after a cooling provocation is a characteristic of the tumour type. Image sequels will be presented for a basal-cell carcinoma and a dysplastic nevus. Since the imaging sequel is acquired with an IR camera equipped with macro lenses, patient motion correction is a prerequisite. This registration process is performed with the help of fiducial markers attached to the patient. Point landmarks are extracted from the markers via Hough transform.