A NEW study presented at the European Society of Radiology Congress (ECR) 2025 in Vienna, Austria, has found that an AI-based SuperResolution reconstruction method for lumbar spine MRI scans significantly reduces scan times without compromising diagnostic quality.
The retrospective study, involving 25 patients, was conducted using a 1.5T MRI scanner (Philips Ingenia 1.5T, Best, NL). The imaging protocol included sagittal and axial sequences in both standard and low resolution, with the AI-based SuperResolution method (SuperRes-AI) compared against the clinical standard Compressed SENSE (CS) technique.
Four experienced readers (two radiologists and two orthopaedic surgeons) evaluated the images for key spinal pathologies, including bone marrow oedema, neuroforaminal stenosis, and disc herniation.
Results showed that the SuperRes-AI method reduced scan times by 31%, cutting acquisition time from 11 minutes and 5 seconds to just 7 minutes and 37 seconds. Crucially, the AI-based approach did not affect diagnostic accuracy. A statistical analysis revealed no significant differences in sensitivity for detecting bone marrow oedema across reader groups and reconstruction methods (p>0.99).
However, the AI-powered reconstruction significantly improved sensitivity for detecting neuroforaminal stenosis among radiologists (p=0.001), highlighting its potential advantage over conventional imaging techniques.
Researchers concluded that AI-based SuperResolution reconstruction allows for faster lumbar spine MRI scans while maintaining high diagnostic standards. This advancement could enhance patient comfort, streamline clinical workflows, and improve pathology detection, particularly for conditions such as neuroforaminal stenosis.
The findings suggest that integrating AI in MRI imaging could be a valuable step forward in medical diagnostics, providing both efficiency and accuracy benefits for healthcare providers and patients alike.
Reference
Hahnfeldt R et al. AI-Driven SuperResolution reconstruction for high-quality, fast MR imaging of the lumbar spine: enhanced image clarity for pathology detection. RPS 110. ECR , 26 February–2 March, 2025.
