Artificial Intelligence revolutionizes skin cancer detection

High-tech meets specialist expertise

Moleanalyzer pro offers physicians the possibility to confirm their diagnosis with various evaluation techniques. For the first time, it is possible to combine specialist expertise with AI and additionally to receive a second opinion – per mouse click – from renowned international skin cancer experts.

AI in dermoscopy

Moleanalyzer pro works with deep learning: The human ability to learn from examples and experiences was transferred to the computer. For this purpose the “Convolutional Neural Network” (CNN) was trained with the currently largest data collection of dermoscopic images including corresponding diagnosis. Due to many years of valuable cooperation with physicians worldwide, the continuous "feeding" of the algorithm succeeds. With growing experience and its own autonomous rules, it is able to distinguish between benign and malignant lesions. The result is a score that supports the risk assessment of both melanocytic and non-melanocytic skin lesions. Shortly, this AI Score will be available for doctors also on mobile devices.

Validated, precise, self-learning

According to the representative study "Man against machine"*, the deep learning algorithm showed an impressively high sensitivity by correctly identifying 95% of malignant skin tumors. In the comparison group, the experts – 58 dermatologists from 17 nations – identified 86.6% of the lesions as malignant. The algorithm also showed a reliably high specificity by identifying 82.5% of benign nevi correctly, while the experts identified 71.3% as benign.

Artificial Intelligence meets human experience

As fascinating as AI is, it cannot replace human experience in the matter of skin cancer. In the end, the doctor decides what to do. In case of doubt, Moleanalyzer pro offers a second opinion service from internationally renowned skin cancer specialists to confirm the diagnosis.

* "Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists", by H.A. Haenssle et al. Annals of Oncology. doi:10.1093/annonc/mdy166