The expectations are high: Artificial Intelligence (AI) has been promising for some time to be the next great revolution in medicine. Even aware of this revolution, the tools based on this concept are gaining step step by step in the area of medical sciences. One of the most recent steps has been given by a team of researchers in Malaga.
Improve images. The step in question has to do with the techniques used to take medical images. The new tool Improve automatic segmentation Of medical images, which in turn allows us to reduce errors in the analysis of complex images, says the team itself.
Segmentation. But what exactly is that automatic medical image segmentation? According to Explain the team itselfmedical segmentation consists in “delimiting anatomical structures or lesions in medical images.”
Thus it is possible not only to diagnose certain diseases more effectively, but also to more easily locate injuries and improve precision in certain treatments. However, this task is not always easy due to the limitations of the techniques used to take these images, such as magnetic resonances and tomographs.
“The precision in the segmentation of medical images is essential to improve diagnoses and treatments,” detailed in a press release Ezequiel López Rubio, who led the study. Our approach demonstrates that it is possible to optimize the reliability of AI models by applying trial time increase techniques, which opens new possibilities in the field of computational medical image. ”
Medsam. For the development of the tool, the team started from a New Model of AI called Medsam, a model that combined with the TTA technique (Test-Time Augmentation). This combination allows, the team itself points out, introduce small transformations in the images at the time of being analyzed.
Thus it is possible to achieve a “more robust system against variations”, with better precision in segmented contours.
Promising results. With Regarding the resultsThe team stressed that the new tool “significantly improves precision” if we compare it with traditional segmentation methods, achieving greater reliability. This greater reliability would be key in images with noise or with little quality, something that, stand out, is common in clinical practice.
Another relevant point for those who have formulated this new tool is flexibility of the methodology and its ability to adapt to different image taking technologies. The details of the study were published In an article In the magazine Mathematics.
{“videoid”: “x8px49v”, “autoplay”: fals, “title”: “Antibiotics are no longer effective and the problem is superbacteria”, “Tag”: “Webedia-prod”, “Duration”: “327”}
The new border. Artificial intelligence continues to see rapid expansion. An expansion that It covers medical disciplines. A good example of this are the diagnostic tools, whose precision would have managed to overcome the “clinical eye” of health personnel, according to They hold their developers.
We still cannot trust our health at AI, but of course the potential of these tools when assisting health professionals and patients themselves grows day by day. These tools are a promising future, but at least for now they are just that.
In Xataka | Surgeons are facing a new challenge: patients who want to resemble their double created by AI
Image | Accuray
(Function () {Window._js_modules = Window._js_modules || {}; var headelement = document.getelegsbytagname (‘head’) (0); if (_js_modules.instagram) {var instagramscript = Document.Createlement (‘script’); }}) ();
It was originally posted in
Xataka
GIPHY App Key not set. Please check settings