Pressemitteilungen Anzeige  Read press releases  Communiqué de presse visualisation  Comunicato stampa  Leer de nota de presa   Link to the US-page
    
Slogan FotoFinder

FotoFinder
in the press

Artificial Intelligence in skin cancer detection

Self-learning algorithm supports dermatologists. The use of artificial intelligence in dermatology practices is no longer talk of the future. With the new expert software FotoFinder Moleanalyzer pro, dermatologists are now able to evaluate moles using an Artificial Intelligence Score (AI Score). First validation results confirm the high accuracy of the algorithm.

Deep Learning – neural networks that learn like the human brain

People learn from examples and experiences. In Artificial Intelligence and Deep Learning, a sub-discipline of AI, this ability has been transferred to neural networks. Like an infant, constantly processing information, the network is confronted with patterns, images and information, and learns autonomously how to categorize them. The larger the amount of training data, the higher is the accuracy.

Assessment of melanocytic and non-melanocytic skin lesions

The same principle was applied to the deep learning algorithm of FotoFinder Moleanalyzer pro. With growing experience and its own autonomous rules, it is able to distinguish between benign and malignant lesions. For this purpose, the algorithm has been trained with a huge number of dermoscopic images and the corresponding diagnoses. The data were obtained from worldwide partnerships with physicians and continue to grow. The result is a score that supports doctors in the risk assessment of both melanocytic and non-melanocytic skin lesions. It is the doctor, however, who ultimately makes the diagnosis. Thus, high-tech and the physician’s expertise are combined.

Validated, precise, self-learning

According to a representative validation at Heidelberg University Clinic, Germany, by Prof. Dr. Holger Haenssle and Dr. Christine Fink, the algorithm is characterized by its high performance and user-friendliness. In a test with 300 sample images, differentiating between benign melanocytic nevi and malignant cutaneous melanomas, it achieved a sensitivity of 96.7 % with a specificity of 85.4 %. The results of the related reader study evaluating the AI Score will be published this year.

Various analysis and evaluation functions

Besides the optional AI score, FotoFinder Moleanalyzer pro offers physicians the possibility to secure their own diagnosis with different techniques. Thus, lesions can be evaluated according to the 3-point checklist, the 7-point checklist or the ABCD rule. In addition, it is possible to request via software a second opinion by renowned international skin cancer experts. The expert system also calculates and compares the size, perimeter and diameter of moles and quantifies their changes in percent.

You want more?

Contact our Press Office by phone at +49 (0) 8563 977200 or via the contact form.

FotoFinder   >  Press
#