Cancer patient. Picture: REUTERS/JIM BOURG
Cancer patient. Picture: REUTERS/JIM BOURG

A Toulouse-based start-up company called Pixience is using artificial intelligence to try to improve the diagnosis of skin cancer, drawing on machine learning to develop a device that it says could help healthcare professionals prioritise urgent cases.

Like many countries around the world, including SA, France is grappling with a shortage of dermatologists. Patients currently wait on average 64 days for a consultation, according to Pixience’s Alexandre Delalleau.

Pixience is developing a device that aims to quickly and reliably gauge whether patients have benign moles or malignant melanomas. If successful, a potential application would be to use it to screen patients who suspected they had skin cancer, and ensure the most urgent cases were placed at the front of the queue, said Delalleau.

"Our aim is to have high-quality screening available in pharmacies that will provide a severity grade for the lesion or mole and then refer the patient to the dermatologist, urgently or not. We don’t want to replace dermatologists — it is a tool to facilitate access [to them]," he said.

"Skin cancer diagnosis is expensive, time-consuming and suboptimal. We think artificial intelligence could improve the sensitivity and specificity of diagnosis, improve the confidence of dermatologists and improve the estimation of excision margins. Today in many cases, lesions have to be removed twice, at extra cost and trouble for the patient," he said.

Delalleau said 86% of excised moles were actually benign, and the average excision cost per mole was more than €1,600.