CNN til diagnosticering af hudkræft

Advisors & Mentors


This project aims to examine the advantages and disadvantages of using artificial intelligence for diagnosing malignant melanoma, as well as a discussion of why it is not yet implemented in healthcare. The structure of our analysis is largely dictated by the TRIN-model. We start off by presenting studies that compare the percentage of correct diagnoses made by experienced dermatologists and a convolutional neural network. By doing this, we find that the CNN performs better than the large majority of dermatologists, and 4% better than the average dermatologist.