Evaluation of Image Quality and Scan Time Efficiency in Accelerated 3D T1-Weighted Pediatric Brain MRI Using Deep...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Pediatric brain MRI protocols often involve long scan times, which can be challenging for young patients.
• Accelerated imaging techniques are crucial for improving patient comfort and reducing motion artifacts.
• Deep learning (DL) algorithms show promise in reconstructing high-quality images from undersampled data.

Purpose of the Study:

• To evaluate the impact of an accelerated 3D T1-weighted pediatric brain MRI protocol with DL-based reconstruction.
• To assess the effects on scan time and diagnostic image quality compared to conventional methods.

Main Methods:

• Retrospective analysis of 46 pediatric patients undergoing conventional and accelerated 3D T1-weighted brain MRI at 3T.
• Quantitative assessment of image quality using coefficient of variation, relative contrast, aSNR, and aCNR.
• Qualitative assessment by radiologists evaluating overall quality, artifacts, gray-white matter differentiation, and lesion conspicuity.

Main Results:

• Accelerated protocol reduced scan times by 29.3% (pre-contrast) and 40.7% (post-contrast).
• DL-based reconstruction significantly improved aSNR, aCNR, and overall image quality, while reducing artifacts (P < 0.05).
• Lesion conspicuity remained comparable between conventional and accelerated DL-reconstructed scans.

Conclusions:

• Deep learning-based reconstruction in accelerated 3D T1-weighted pediatric brain MRI is effective.
• This approach significantly shortens acquisition time and enhances image quality.
• It presents a viable and beneficial option for pediatric neuroimaging.