The start-up Therapixel raises €5m to improve breast cancer detection

Date:
Changed on 08/10/2020
Specialising in Artificial Intelligence applied to medical imaging, Therapixel, an Inria spin-off based in Sophia Antipolis, has just raised 5 million euros in a Series A round (optimisation phase) in order to develop its artificial intelligence system used to help diagnose breast cancer.
Illustration mammographie
© Unsplash / Photo Victoria Strukovskaya

 

Set up in 2013 by Olivier Clatz (Inria Sophia Antipolis - Méditerranée) and Pierre Fillard (Inria Saclay Ile-de-France), Therapixel has emerged as one of the leading start-ups specialising in artificial intelligence applied to medical imaging. Its algorithm interprets mammograms, enabling highly reliable breast cancer detection at first reading.

 

The primary objective is to help with first reading, explains Pierre Fillard. In other words, to assist radiologists without replacing them. The company’s software, Mammo Screen, helps ensure that radiologists make the right decision. The algorithm that it is developing acts as a virtual assistant, capable of detecting the first signs of cancer at an earlier stage.

 

 

One in eight women are affected by breast cancer at some point in their lives, with 250 million mammograms analysed globally each year. A mammogram may contain up to 3,000 images for analysis, and is viewed several times: the first doctor to view it will begin by detecting and assessing any potential abnormalities, before a second either confirms or contradicts the original analysis. The automatic learning algorithm developed by Therapixel is introduced during the first stage, the goal being to speed up the process while making it more precise. Breast cancer diagnosis is very challenging with only 0.5% of mammograms containing cancerous tissue and 10% are false positive.

A successful challenge

Prior to its work in the field of breast cancer detection, the start-up began life by developing Fluid, medical imaging software used in operating theatres that is controlled remotely using motion sensors. This tool allows surgeons to consult x-rays and to navigate within the interface remotely using gestures. This prevents their gloves from being contaminated by computing equipment, and saves time into the bargain. The product, which is already on the market, enabled the company to raise 600,000 euros in seed money in 2015.

The pivotal moment came in 2017 when the start-up won the DREAM Digital Mammography Challenge, an international competition that pits almost 1,200 research teams against each other as they attempt to devise algorithms for use in mammogram diagnostic assistance. What made this such a challenging undertaking was the vast quantity of data to be processed (640,000 anonymous digital mammogram images, taken from more than 86,000 patients) coupled with extremely limited computing resources (14 days in the Amazon cloud).

Attract the best talent and maintain technological leadership

This first fundraiser was a significant turning point in the Therapixel story. Pierre Fillard, co-founder and formerly technical director, has since taken on the role of managing director. In doing so, he replaced Olivier Clatz, who was named director of the “AI and diagnosis” programme by the Innovation Board (French Gvt.) on 1st March 2019.

On the back of this major financial contribution, Therapixel is also keen to strengthen its corporate governance and to expand its board of directors. Sacha Loiseau, founder and CEO of Mauna Kea Technologies, joined the board of directors as an independent member and chair. Maximilien Fournier-Sourdille, business manager at Omnes and an experienced investor in the MedTech sector, also joined the board of administrators, alongside Yoann Bonnamour from M Capital Partners.

Therapixel's research teams operate out of Paris, while the software development side is based in Sophia Antipolis. They intend to exploit these new resources in order to attract the very best talent and to maintain their technological lead.

Therapixel
© Therapixel / Photo Ari Rossner