BACKGROUND

The last 30 years show a 4-fold increased melanoma incidence. 25% of malignant melanomas are not detected in time and can lead to death.

Systems for the diagnosis of melanoma from images of naevi obtained under controlled conditions have shown comparable efficiency with dermatologists’ diagnosis. However, their robustness to analyze daily routine images has been discutable.

SkinAn, “the Skin Analyzer”, offers automatic melanoma diagnosis based on the analysis of images of pigmented skin lesions obtained under uncontrolled conditions.

HOW IT WORKS

Computer-aided diagnosis for melanoma detection by clinicians. Innovative concept that combines: i) a digital image capture device, ii) an image processing software, iii) a reference database of over 3000 naevi pictures (benign, atypical, melanoma) previously assessed by 12 senior dermatologists (contour, appearance, color...).

KEY BENEFITS vs. STATE OF THE ART

SkinAn out-performs the ABCDEs rules currently in use:

• Easy, early and reliable naevi detection using malignancy-predictive features,
• Non-invasive and economical system,
• Real-time and mobile monitoring of naevi,
• Classifier combining image-processing with machine-learning techniques,
• Website for interactive exchanges between patients and clinicians

DEVELOPMENT STATUS

• Passed operational testing phase on DEGETEL platform
• Classifier shows comparable performances with respect to dermatologists (95% sensitivity, 60% specificity)
• Demonstrator available

APPLICATIONS

Remote diagnosis of melanoma skin cancer before medical diagnosis