Description: In this project we will investigate new approaches for skin cancer screening using automated image analysis on clinical photography, dermoscopy and reflectance confocal microscopy data sources. Skin cancer is the most common and costly cancer in Australia (AIHW 2012). Early detection and treatment of skin cancer is vital for good patient outcomes. Thus, we hypothesize that significant benefits can be realised by combining clinical assessment with objective image analysis developing a decision support system for the early recognition of skin cancer.
Our approach is focused on automated, objective analysis of three non-invasive imaging modalities: total body clinical photography, dermoscopy and reflectance confocal microscopy. We are applying imaging methodologies to a large group of patients, with a focus on recording information about all lesions present on the patient, rather than just the 'suspicious' lesions. The automated analysis outcomes will then be validated with expert clinical and pathological assessment.
Aims: To develop and validate automated image analysis approaches for the early detection of skin cancer.
Status: We have showed successful identification and quantification of actinic keratosis lesions using automated analysis of digital camera images and videos of volunteer skin. We are now developing and validating a texture analysis tool to segment the skin strata in normal and photodamaged skin using reflectance confocal microscopy data.