Fractures: better identify them thanks to artificial intelligence

Fractures better identify them thanks to artificial intelligence

  • News
  • Published ,


    Reading 1 min.

    According to a study conducted by a professor of radiology at Boston University in the United States, artificial intelligence would be able to detect fractures that are difficult to see with the naked eye. This will further improve the quality of diagnoses made by doctors.

    The algorithm put in place would automatically and very quickly detect fractures of the limbs, pelvis, torso, lumbar spine and rib cage. Faced with the expertise of specialists, it has thus made it possible to reduce missed fractures by 29% while improving their specificity by 5%.

    Faster diagnostics

    With this algorithm, diagnoses could be rendered more reliably and above all faster than human interpretation. The first benefit of such technology would be to reduce waiting times in hospitals, especially in emergency departments. According to those responsible for this study, this work concerning the diagnosis of fractures could very well be applied in the same way to various diseases. The idea here is not to replace doctors but to assist them intelligently in their work.

    AI for breast cancer too

    Already in 2020, Google researchers had published a study showing that their algorithm was more effective in detecting certain breast cancers, compared to radiologists who were responsible for reading the same mammograms. But the system is still not perfect. Because in some special cases, the correct diagnosis has been made only by radiologists. Radiologists are not going to disappear in favor of machines, therefore, as Bernard Nordlinger, member of the Academy of Medicine, points out to our colleagues from Sciences et Avenir. “AI will help us but in no way replace us. It’s a quick and logical method, but it doesn’t make any sense, except that it’s the first quality of any doctor. AI remains a diagnostic aid“. And this, even in cases where she does better than the man.

    dts7