The client, a dermatology department of an important hospital, wanted to have a tool able to identify three nail diseases (melanonychia, dystrophy and onychomychosis) and, most importantly, be able to differentiate between healthy and un healthy nails so as to assess, automatically and quickly, whether a patient needed to make an appointment with the dermatologist.
Sigma’s Computer Vision technology can recognize patterns with a very high accuracy, which makes it ideal for this use case.
The dermatology clinic had a large data base of annotated nail images (healthy and unhealthy), which allowed to adapt Sigma’s Computer Vision technology to the detection and identification of nail diseases very quickly and easily. Sigma’s team preprocessed the images to reduce variability and optimize the result of the machine learning process, and fine tuned the technology parameters for this type of images.
Sigma also developed a user interface to visualize the results in a way that the dermatologist could rapidly confirm the diagnosis of the disease.
The resulting tool was able to identify healthy nails with an accuracy of 95%, and to detect and identify the three nail diseases with the following accuracies:
- Onychomicosis: 83%
- Dystrophy: 88%
- Melanonychia: 94%.