Researchers have developed a deep-learning platform to speed up the MRI reconstruction process. The framework takes a fraction of the traditional technique’s measurements. It uses plug-and-play algorithms to combine physics-driven data acquisition models with state-of-the-art learned image models. The plug-and-play techniques recover pictures faster, with improved quality and potentially superior diagnostic utility than existing MRI reconstruction methods. The researchers also used deep learning-based denoisers to refine the images further.
The plug-and-play solution is distinct in that it combines machine learning-based approaches with traditional engineering physics approaches, iterating between them to deliver the image.
Once fully implemented, the methodology could cut MRI reconstruction time in half. It requires no hardware modification; a computer workstation connected to the MRI scanner will perform the computations. Clinicians want to take a look at the MRI images within seconds. After development, the algorithm can recover the scan images in a matter of seconds, almost in real-time.
To demonstrate its broad applicability, the team will validate its framework using MRI data from pediatric and adult patients, specifically for cardiac cine and brain imaging. If successful, the acceleration and image quality improvement afforded by these developments will benefit almost all MRI applications.
Making MRIs faster is especially critical for pediatric patients. Pediatric imaging is an even bigger issue because smaller kids don’t stay still in the MRI scanners, so they have to sedate them. Growing literature says sedation has negative long-term consequences, so we need to minimize it.
Related Content: AI-Based Dynamic Brain Imaging Technology