Introduction to AI in Skin Cancer Detection
Recent advancements in technology have introduced artificial intelligence (AI) as a promising tool in the early detection of skin cancers. While AI shows significant sensitivity in identifying potential cancerous lesions, it still lacks the specificity required to replace human expertise. This was the conclusion of a study presented by Dr. Alexandre Lellouch at the Medintechs conference in Paris, which evaluated a device combining AI with dermatoscopy.
The Decline in Dermatologists and Rising Skin Cancer Cases
Over the past decade, the number of dermatologists has decreased by nearly 25%, while cases of skin cancers, including carcinomas and melanomas, continue to rise. This demographic shift poses a challenge in ensuring timely diagnosis and treatment, which are crucial for patient outcomes. To address this issue, Dr. Alexandre Lellouch, a plastic surgeon and researcher at Harvard University and Cedars Sinai in the United States, along with Dr. Moshe Assouline, a general practitioner in Paris, and a team of dermatologists, assessed the performance of an AI-dermatoscopy device for skin cancer diagnosis. Preliminary results were shared at the Medintechs conference, with a full publication pending in a specialized journal. These findings suggest that while AI can aid in screening, it cannot replace the expertise of a medical professional.
Study Overview and Methodology
The study, conducted at Maison Abeille in Paris, aimed to evaluate a new skin cancer diagnostic strategy inspired by an Australian model. “In this study, we share our experience after two years of using the FotoFinder imaging device coupled with AI-assisted dermatoscopy,” explained Dr. Lellouch. The research involved a multidisciplinary team of general practitioners, dermatologists, and plastic surgeons. The trial compared AI-generated results with clinical evaluations by dermatologists or general practitioners trained in dermatoscopy.
How FotoFinder Technology Works
FotoFinder is a state-of-the-art medical imaging device primarily used in dermatology for early skin cancer detection and monitoring existing skin lesions. It allows for a precise and automated high-resolution mapping of the entire skin surface in minutes. The associated system, known as Bodyscan Master, identifies and analyzes skin lesions, providing detailed visualization of each skin element. With magnification lenses ranging from 20 to 70 times, FotoFinder can analyze lesions as small as 5 millimeters.
The examination begins with an automated full-body photograph, followed by digital dermatoscopy to obtain detailed images of each lesion from various angles. AI then assigns a risk score to each mole or lesion based on classic criteria such as size, color, asymmetry, and evolution, following the ABCDE rule. A supervising physician verifies the alignment between the AI results and the clinical examination.
Study Results and Implications
The study analyzed 403 lesions in 271 patients examined by FotoFinder between November 2022 and July 2024 at Maison Abeille. A total of 130 skin cancers were detected. Preliminary results indicate that the AI-dermatoscopy technology has a high sensitivity (92.31%) for detecting malignant lesions. However, its specificity is still low (40.80%), posing a risk of overdiagnosis. “While AI identified malignant lesions that clinical evaluation alone might have missed, it also misclassified 8 malignant lesions as benign,” acknowledged Dr. Lellouch. These findings highlight the importance of a hybrid model where AI assists but does not replace the physician’s expertise.
Future Perspectives and Personalized Medicine
Additionally, images and analysis results are stored in a database for comparison in future examinations. Currently, AI diagnoses must be corroborated by a clinical diagnosis from a physician. “These new technologies open up future possibilities,” stated Dr. Lellouch. “Interactions between doctors and AI will increasingly lead to more personalized and effective medicine.”
For more updates, subscribe to Medscape newsletters and follow Medscape’s French edition on Bluesky.
**Source:** https://francais.medscape.com/voirarticle/3612727