A recent exciting breakthrough by researchers at MIT has revolutionized the use of technology in minimally invasive surgeries. This novel technique, which uses an advanced form of X-ray matching, increases the precision with which surgeons can navigate their tools, making procedures faster, safer and potentially more accessible to a broader population.
Minimally invasive surgeries – procedures that enter the body through tiny incisions – rely heavily on real-time X-rays for guidance. The nature of these X-rays, however, can make it challenging to pinpoint the exact location and orientation of surgical instruments. In turn, this problem can lead to complications. Clinicians usually manually align these X-rays with preoperative 3D scans like CTs or MRIs to increase the accuracy. While there are AI tools designed to assist in this process, the reality is they often fall short, struggling to consistently achieve an accurate match between X-rays and 3D images across different patients.
However, the team at MIT, with their collaborators, have developed a ground-breaking system named XVR, an innovative application of AI that is adapted uniquely to each patient in about five minutes. What does this mean in practice? The AI model can then match a patient’s X-rays with their 3D scans within seconds and with remarkable precision. The implementation of xvr has significantly outperformed existing AI methods — quite a feat given the complexity and diversity of patient demographics and medical procedures.
The potential impact of xvr is far-reaching, as it could increase the accessibility of life-saving procedures for many Americans who live more than an hour away from medical facilities capable of performing noninvasive procedures. This is a particularly promising prospect in emergencies situation requiring immediate treatment, such as stroke interventions.
One detail setting xvr apart is how it’s developed on a patient-specific basis, instead of the traditional one-size-fits-all approach, often used by other AI models. xvr generates thousands of synthetic X-rays from a patient’s preoperative 3D scan, using a physics-based simulation to ensure realism. This approach nixes the risk of errors often associated with generative AI methods. Leveraging data from over 2,000 whole-body 3D scans, xvr can perform 2D/3D registration quickly and accurately.
Looking ahead, the researchers plan to enhance the real-time deployment of xvr and explore its usefulness in more complex scenarios, such as moving body parts. Collaborations are already underway with surgical robotics companies and clinical groups to evolve this research into practical surgical navigation tools. This game-changing technology is supported by several institutions, like the National Institutes of Health (NIH), the MIT CSAIL-Wistron Program, and the MIT-IBM Computing Research Lab.
For more information on AI automation solutions, consider visiting implementi.ai. The original news release can be found at MIT News.
This website uses cookies.