Deep-Learning Accelerated Vessel Wall Imaging Using T1-SPACE at Ultra-High-Field Strength MRI

Area of Science:

• Neuroimaging
• Radiology
• Medical Physics

Background:

• Intracranial vessel wall imaging (IC-VWI) at 7T is crucial for diagnosing cerebrovascular diseases.
• Standard-of-care (SOC) T1-SPACE sequences have limitations in image quality and acquisition time.
• Deep learning-based image reconstruction (DLBIR) offers potential for accelerated and improved imaging.

Purpose of the Study:

• To assess the feasibility and performance of DLBIR-accelerated T1-SPACE for 7T IC-VWI.
• To compare image quality, artifacts, and scan time against SOC T1-SPACE.
• To evaluate the clinical utility of DLBIR in IC-VWI.

Main Methods:

• Retrospective study of 36 patients undergoing 7T IC-VWI with both SOC T1-SPACE and DLBIR-T1-SPACE sequences.
• Two neuroradiologists assessed image quality (noise, artifacts, sharpness, overall quality, wall/lumen visualization) using a 4-point Likert scale.
• Statistical analysis included cumulative-logit mixed-effects models, Wilcoxon tests, and Bland-Altman analysis.

Main Results:

• DLBIR-T1-SPACE showed significantly higher ratings for wall and lumen visualization compared to SOC T1-SPACE (p<0.001).
• DLBIR-T1-SPACE demonstrated superior image quality across noise, artifacts, and sharpness (p<0.001).
• Acquisition time was reduced by 20% with DLBIR-T1-SPACE (6:00 min vs. 7:30 min).

Conclusions:

• DLBIR-accelerated T1-SPACE is a feasible and effective technique for 7T IC-VWI.
• DLBIR significantly enhances image quality and reduces scan time, offering potential clinical benefits.
• This method holds promise for improved diagnosis and management of intracranial vascular pathologies.