This project aimed to design and effectively develop and implement a framework of risk triage of skin cancer, which uses a new generation of mobile devices in its architecture to capture the images. The framework automatically pre-processes and segments mobile-acquired skin moles images, as well as perform extraction of significant features for risk assessment and Melanoma pre-diagnosis purposes. The automatic risk assessment is based on machine learning methods using extracted features, additional information available submitted by the patient and an adaptive reference atlas of classified skin lesions. The images that make up the reference atlas are previously classified by dermatologists and are used to provide a highly reliable triage of lesions based on images acquired by low cost devices such as smartphones.
INEGI-LAETA (coordinator); Instituto Português de Oncologia (IPO).
> Developed pre-processing computational techniques for image enhancement and illuminance corrections of mobile-acquired images;
> Developed and implemented a segmentation method specifically designed for skin lesion images acquired from mobile devices;
> Selected and effective extracted significant features from mobile-acquired skin images for risk assessment purposes;
> Developed machine learning approaches for the automatic classification of skin lesions. The machine learning classifiers were trained with an adaptive reference atlas of skin lesion images;
> Designed and implemented a Mobile Risk Triage Framework prototype, which automatically pre-processes, segments and extracts significant features for skin cancer pre-diagnosis and risk assessment of mobile-acquired skin images.
“SMARTSKINS: A Novel Framework for Supervised Mobile Assessment and Risk Triage of Skin lesions via Non-invasive Screening” is a joint project with INEGI-LAETA, Fraunhofer AICOS and IPO-Porto financially supported by Fundação para a Ciência e a Tecnologia in Portugal (PTDC/BBB-BMD/3088/2012).
For any additional information, please contact us using the inquiries form.