3D magnetization-prepared rapid gradient echo with a RF saturation band on a compact 3T scanner.

Background: Flow-related ghost artifact from a 3D Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequence results from unsaturated magnetization of incoming arterial flow. This is especially common on MR scanners equipped with smaller radiofrequency (RF) transmit coils. A high-performance compact 3T (C3T) scanner features a smaller RF-transmit coil (inner diameter 37 cm, length 40 cm), leading to a rapid fall-off of the B1 field below the neck. This configuration results in more intense flow-related ghost artifacts, especially in younger patients.

Purpose: The C3T scanner's smaller RF-transmit coil provides RF field coverage over the brain region, causing bright in-flow arterial signals and flow-related artifacts in the conventional 3D-MPRAGE scans. The purpose of this study is to suppress these artifacts by adding a RF saturation band (RFSB) pulses to the 3D-MPRAGE sequence.

Methods: The RFSB was added to the 3D-MPRAGE sequence as a preparation pulse option. To test the effectiveness, 37 subjects were scanned on the C3T under an IRB-approved protocol using 3D-MPRAGE with and without RFSB. Ten of those subjects underwent repeated scans with and without RFSB to evaluate test-retest reliability. A consensus evaluation by two neuroradiologists was performed on all data to compare signal to noise ratio, image contrast, presence of artifacts, and diagnostic confidence. Quantitative analysis included calculating test-retest differences by image subtraction and evaluating the variance in flow-artifact-induced image intensity between the scans with and without RFSB. Additionally, one subject was scanned on a whole-body 3T scanner using a transmit/receive (T/R) head coil to demonstrate the method's applicability across different MRI platforms as a proof of concept.

Results: The Wilcoxon signed-rank test of the neuroradiologist evaluations showed a significant reduction in artifacts and an improvement in diagnostic confidence in the posterior fossa region with and without RFSB (p < 0.0001). Test-retest analysis showed that adding RFSB significantly reduced image intensity variability in the cerebellum, even among subjects without visible flow artifacts. The normalized difference decreased from 8.71% to 6.41% (p = 0.0059), suggesting improved image reliability in regions prone to flow-related artifacts. Additionally, similar findings were observed in scans on a whole-body 3T with a T/R head coil, demonstrating the broader applicability of this method.

Conclusion: Incorporating RFSB into 3D-MPRAGE scans effectively reduces flow-related ghost artifact on the C3T scanner, improving image quality and diagnostic confidence. These findings suggest that the proposed method could be widely implemented across MRI systems utilizing a smaller RF-transmit coil.