Engineers at MIT created an artificial intelligence tool that can help detect skin cancer. Melanoma is a type malignant tumor responsible for more than 70% of all skin cancer-related deaths worldwide. For years, physicians have relied on visual inspection to identify suspicious pigmented lesions (SPLs).
SPLs could be indication of skin cancer. Early-stage identification of SPLs can improve melanoma prognosis and scientificantly reduce treatment cost. Researchers face difficulties when trying to quickly find and priorities SPLs. This is mainly due to the high volume of pigmented lesions that often need to be evaluated for potential biopsies.
Researchers from MIT presented a new AI mechanism that uses deep convolutional neural networks (DNCCs). There are used to analyze SPLs through world-field photography widely available in most smartphones and personal cameras. The image shows large skin sections from a patient. Then an automated system detects, extracts, and analyzes all pigmented skin lesions observable in the image.
The pre-trained convolutional neural networks (DNCC) determines the suspiciousness of individual pigmented lesions and the possibility of signalling early-stage melanoma. Such a method could help medical system to provide comprehensive skin screening at scale. Using AI, the researchers trained the system using 20,388 world field images from 133 patients.
Dermatologists working with researchers visually classified the lesions in the images for comparison. They found that the system achieved more than 90.3% helping them avoid a difficult and time-consuming task. Researchers concluded computer vision and deep neural networks achieve highly accurate results that are comparable to expert dermatologists.