Breast cancers are very frequent and hard to detect 1 in 8 women will have breast cancer: it is both the most common and the most fatal female cancer. Detected early, current management has achieved a five-year survival rate of 99%. The basic screening test, mammography, is a real interpretive issue for radiologists. The difficulty is such that in some countries such as France, each image is read by 2 physicians. Out of every 1000 women screened, on average only 5 will have breast cancer. But 100 will be called for further testing. Trained radiologists are rare. Hospitals around the world are facing a chronic shortage of specialists to perform and analyse mammograms. There is a concern that breast cancer screening programs may not be able to cope with the increasing number of women who will need mammograms in the coming years.
Therapixel is developing Artificial Intelligence(AI)to analyse screening mammograms. Based on cases interpreted by expert radiologists and using deep learning techniques, algorithms learn to detect suspicious areas. Using databases of several thousand examinations, the algorithm is now reaching radiologists' performance*. Combining the expertise of radiologists and the power of these algorithms makes it possible to detect breast cancer earlier and reduce the need for additional tests. *Ongoing clinical study
By assisting radiologists to identify suspicious areas, cancers are detected earlier.
With the help of AI, radiologists will make a more accurate diagnosis and reduce the number of additional examinations.
Initial results show that 30% of cancers can be detected 1 year earlier with the assistance of Artificial Intelligence
International competition on the detection of cancer in mammography examinations. Dataset of 640k images, organised by: FDA, NCIO, SCA, IBM, Mount Sinai,etc.
Winner of the 018 Innovation Award by SFR, Medicen and the French Association for Medical Device (SNITEM). Most innovative company in medical imaging.
OVH, Europe’s leading provider of cloud hosting organises, an annual competition between the best projects using cloud technology.
Not for Sale in the US