In a study that was published in Scientific Reports, scientists from Addenbrooke’s Hospital, Anglia Ruskin University, Check4Cancer, and the University of Essex used combination theory and machine learning to narrow down 22 clinical features to the seven most crucial ones that could indicate whether or not a skin lesion is suspicious.
Image Credit: Africa Studio/Shutterstock.com
Scientists in the East of England have developed a method for using artificial intelligence to detect skin cancer. In this new study, the AI tool outperformed existing methods.
The AI model was developed by researchers using information from 25,105 patients’ 53,601 skin lesions.
The new characteristics include the hair color at age 15, whether the lesion was pink or inflamed, and whether it has recently changed in size, color, or shape.
Researchers used proportional weighting on these seven features to create the new C4C Risk Score, which has a 69% accuracy. The study outperformed other methods, such as 7PCL (62%) and Williams score (60%).
The older approaches, which only addressed melanoma, a particular kind of skin cancer, did not take into account some of the new risk factors they found, such as lesion age, pinkness, and hair color, which were significant for all forms of skin cancer.
This study shows the importance of using clinical data in skin lesion classification, which should help to improve the detection of skin cancer. Our new AI model, which combines the C4C risk score together with skin lesion images, could lead to a reduction in the need for patient referrals for biopsies, shorter waiting times for skin cancer diagnosis and treatment, and improved outcomes for patients.
Gordon Wishart, Visiting Professor of Cancer Surgery, Anglia Ruskin University
Consultant Plastic Surgeon Per Hall, who recently retired from Addenbrooke’s, added, “The added value that this paper brings is the ability to help identify patients whose skin lesions are suspicious enough to justify onward referral for face-to-face analysis. Emphasis in the past has been on pigmented lesions and melanoma but other things grow on the skin that need sorting out such as basal cell carcinomas and squamous cell carcinomas.”
“The NHS is deluged with referrals for skin lesion analysis – the vast majority are in fact innocent. This work is geared towards sifting out lesions that are potentially serious and identifying those patients whose skin is more prone to developing cancers so they can be seen quickly,” Hall concluded.
A Knowledge Transfer Partnership (KTP) Grant from Innovate UK provided partial funding for the study.
It is anticipated that the AI model will receive regulatory approval by 2025.
Journal Reference:
Islam, S. et. al. (2024) Leveraging AI and patient metadata to develop a novel risk score for skin cancer detection. Scientific Reports. doi.org/10.1038/s41598-024-71244-2