The researchers tested the system against visually classified lesions by dermatologists and found that it achieved over 90% sensitivity in distinguishing SPLs from nonsuspicious lesions, skin, and complex backgrounds.
Malignant melanoma is the major cause of death from skin cancer and is more likely to be reported and accurately diagnosed than non-melanoma skin cancers, according to the World Health Organisation. In 2020, there were more than 3.2 lakh new cases and over 57,000 deaths associated with melanoma of skin across the globe.
A new artificial intelligence (AI) tool, developed by a team of researchers led by the Massachusetts Institute of Technology (MIT), can quickly analyse images of patients’ skin to detect cancer more effectively and efficiently.
The researchers have detailed their work in a paper titled ‘Using Deep Learning for Dermatologist-level Detection of Suspicious Pigmented Skin Lesions from Wide-field Images,’ published in the medical journal Science Translational Medicine.
Suspicious pigmented lesions (SPLs), which can be indicative of skin cancer, are normally spotted by physicians through visual examination. However, quickly finding and prioritising SPLs is difficult, due to the high volume of pigmented lesions that often need to be evaluated for potential biopsies, MIT noted in a release.
“Early detection of SPLs can save lives; however, the current capacity of medical systems to provide comprehensive skin screenings at scale are still lacking,” Luis R. Soenksen, an author of the paper and a Venture Builder in AI and Healthcare at MIT, said in a release.
The SPL analysis system uses wide-field photographs of large regions of patients’ bodies captured through commonly available smartphone and personal cameras. It then processes these images using deep convolutional neural networks (DCNNs) in a timely and effective manner to identify and screen for early-stage melanoma, the institute explained.
The team used AI to train the system with 20,388 wide-field images from 133 patients at the Hospital Gregorio Marañón in Madrid, as well as publicly available images, it added.
The researchers tested the system against visually classified lesions by dermatologists and found that it achieved over 90% sensitivity in distinguishing SPLs from nonsuspicious lesions, skin, and complex backgrounds, MIT noted.