Deep Learning Image Processing Enables 40% Faster Spinal MR Scans Which Match or Exceed Quality of Standard of Care :...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Spinal magnetic resonance imaging (MRI) is crucial for diagnosing neurological conditions.
• Reducing MRI scan times is a significant clinical challenge.
• Deep learning (DL) shows potential for accelerating medical image acquisition and reconstruction.

Purpose of the Study:

• To evaluate the performance of a 40% scan-time reduced spinal MRI protocol reconstructed with DL.
• To compare the image quality and diagnostic integrity of accelerated DL-reconstructed MRI with standard-of-care (SOC) MRI.

Main Methods:

• A prospective multicenter study involving 61 patients.
• Standard of care (SOC) and accelerated (FAST) spine MRI scans were acquired.
• Deep learning (DL) was used to reconstruct the FAST scans (FAST-DL).
• Three neuroradiologists assessed image quality, and quantitative metrics (SSIM, L1) were used to evaluate image integrity.

Main Results:

• FAST-DL was qualitatively superior to SOC for signal-to-noise ratio (SNR) and artifact reduction.
• Image quality for other features was equivalent between FAST-DL and SOC.
• High SSIM values confirmed no significant image corruption from DL processing.

Conclusions:

• Deep learning enables a 40% reduction in spinal MRI scan time.
• Diagnostic integrity and image quality are maintained with the accelerated protocol.
• The FAST-DL approach offers perceived benefits in SNR and artifact reduction, suggesting clinical utility.