Skin Disease Assessment Using Computer Vision
Purpose and goal: The aim of the project was to develop a prototype system which, using computer vision and machine learning, automatically can assess and grade skin diseases, in particular atopic dermatitis using only digital photographs of the patient´s skin. Some preparatory software development was completed, e.g. a mobile app for data collection using mobile phones and a machine learning framework. The project had to be terminated prematurely when it became clear that the training data needed would not be made available within the time frame of the project, mainly due to a high workload at SUS. Expected results and effects: To some degree, the software developed can be reused within other projects. Tests of the prototyped algorithms indicate that the end goal is technically feasible given enough training data. Approach and implementation: In retrospect, the time plan for the project might seem too optimistic. Other projects have shown the difficulty in acquiring training data of different kinds, especially medical due to the regulatory hurdles. Letting the success of the project rely on the access to data was a gamble that did not pay out, but this is exactly what the AI revolution of the last few years is based on. The technology is mostly already in place, what´s missing is access to high-quality training data in large enough quantities.
Fredrik Kahl (contact)
Full Professor at Chalmers, Electrical Engineering, Signal Processing and Biomedical Engineering, Imaging and Image Analysis
Skåne University Hospital
Project ID: 2017-04585
Funding Chalmers participation during 2017–2019
Related Areas of Advance and Infrastructure