Raidium ou comment utiliser l’IA pour soutenir les radiologues

[ad_1]

The popularity of artificial intelligence continues to grow, particularly with the advent of ChatGPT. And the start-up Raidium is riding the wave since it offers to facilitate the work of radiologists. Helping with diagnosis and carrying out assessments, Raidium wants to integrate AI into medicine. Focus on this young company.

In January 2022, Pierre Manceron, graduate of CentraleSupélec and ENS Paris-Saclay, and radiologist Paul Hérent, both former employees of the medical technology unicorn Owkin, founded the start-up Raidium. Their objective ? Using AI to make radiologists’ work easier. The new generation ofartificial intelligence which emerged at the beginning of the 2020s and the attractiveness generated by it allows them to embark on the adventure while taking advantage of the best conditions.

Nowadays, the work of radiologists is becoming more dense, despite certain algorithms of speech Recognition which already lightens the work a little. But with the increase in the speed ofmedical imaging, doctors are forced to study more files per hour than before. The founders of Raidium want to solve this problem by speeding up radiologists’ analysis time using AI models. This would involve directly generating the radiological assessment or detecting anomalies on the images, always under the control of the doctor. In addition, with this time compression, radiologists do not have the opportunity to study certain elements such as the size of organs and tumors. The amount of data to be examined is enormous, as is the mental load on specialists.

Thousands of data to compile

In order to meet its objectives, Raidium uses foundation models. These are models that learn in an unsupervised manner with a quantity of parameters numbering in the billions. These programs are capable of assimilating a very large amount of information by mixing their type. They can thus process images with text. A major advantage in medical imaging since the images are always accompanied by a report. This learning capacity of AI leads to better representation of data and greater finesse in analysis.

After collecting thousands of data from different hospitals, the co-founders will now offer a model of their product, in order to show its versatility. Raidium already has a partnership with Parisian hospitals to tackle liver diseases such as portal hypertension (major complication of cirrhosis) or Nash (non-alcoholic steatohepatitis also called “fatty liver disease”). These pathologies are poorly diagnosed and therefore poorly treated. The product developed by the start-up could also be useful in the diagnosis of certain cancers. When the disease metastasizes, the analyzes are complicated, because the evolution of tumors in the different organs must be studied, something that current AI is not capable of doing today. The AI ​​proposed by Raidium could be used to perform virtual biopsies for Nash, non-invasively predict invasive markers or even better evaluate tumor volume in the context of clinical trials.

[ad_2]

Source link

Scroll to Top