Software helps classify skin cancer
Researchers at the University of Edinburgh (Edinburgh, UK) have developed diagnostic software that can help health care workers to correctly identify different types of skin lesions, leading to more effective diagnosis of skin cancer.
500,000 referrals for potential skin cancer are made from primary care to specialist secondary care per year in the UK. However for every confirmed skin cancer case, 10 to 20 turn out benign. Unnecessary referrals waste valuable resource and time, and also prolong patient anxiety.
The so-called Dermofit software -- which can be delivered on a mobile, tablet or web-based system -- contains more than 5000 high-quality images of skin cancers that have been annotated diagnostically by a team of consultant dermatologists and pathologists.
The organization of the image database enables physicians to navigate through thousands of the images to find a skin lesion comparable to those of a patient in around 60 seconds.
Use of the Dermofit system could help to reduce the numbers of false positive referrals to consultant dermatologists while also flagging up serious cases for faster secondary care follow-up and treatment.
A published proof-of-concept study has shown that novices (students and lay people) using the system outperform students who have received two weeks’ clinical tuition.
Currently, the University of Edinburgh is seeking development partners and/or licensees to commercialize the software.
More information can be found here.
-- by Dave Wilson, Senior Editor, Vision Systems Design