Non-contrast MR angiography (MRA) is a promising method to diagnose and follow-up of vascular diseases such as dissecting aortic aneurysm¹. Typically,3D gated balanced steady-state free precession (bSSFP) has been applied as a conventional non-enhanced MRA for the aorta. Although bSSFP theoretically has high signal-to-noise ratio (SNR) and thus it provides high-contrast MRA with shorter acquisition time, it has several limitations. For example, bSSFP is difficult to separate the arteriovenous signals in some cases due to black-band artifacts related to its sensitivity to magnetic field inhomogeneities. Furthermore, it often suffers from insufficient fat suppression especially when large field-of-view (FOV) is applied.

To overcome these limitations of conventional bSSFP sequence, we propose a new technique based on gradient echo DIXON sequence with flow-independent relaxation-enhanced non-contrast MRA technique (Relaxation-Enhanced Angiography without Contrast and Triggering: REACT). The purpose of this study was to optimize the imaging parameters for REACT and to evaluate the feasibility by comparison of conventional sequences.

REACT sequence

REACT sequence consists of improved mDIXON gradient echo sequence (mDIXON XD), non-volume-selective short tau inversion recovery (STIR) pulse and T2 preparation (T2prep) prep-pulse (Fig.1). First, mDIXON XD provides robust fat suppression even if the large FOV is applied (Fig.1a). Besides, STIR pulse is applied for increasing the contrast between blood vessels and background signals such as muscle². Moreover, the T2prep pulse is adopted to differentiate between arteries and veins by utilizing the differences of their T2 relaxation times³. Consequently, REACT would provide flow-independent relaxation-based MRA while keeping high robustness for uniform fat suppression at entire large FOV.

Experiments

A total of six volunteers were examined on 3.0T whole-body clinical systems (Ingenia, Philips Healthcare). The study was approved by the local IRB, and written informed consent was obtained from all subjects.

(1) Parameter optimization: Theoretically, arteriovenous contrast depends on the preparation time (prep-time) of T2prep pulse as aforementioned. Hence, we investigated the optimal prep-time by changing the prep-times (none, 30, 50, 70, and 90ms). To quantitatively compare the effect of respective prep-times, we measured the contrast-ratio (CR) between aortic artery and inferior vena cava (IVC) (CR_Artery-Vein) and between artery and muscle (CR_{Artery-Muscle}), respectively. Also images of respective prep-time were visually assessed and compared by the radiologist and technologists.

(2) Comparison with conventional bSSFP sequence: To demonstrate the feasibility of REACT sequence, we compared the CR between aortic artery and IVC with conventional bSSFP sequence. Subsequently, we compared visually the image quality with conventional bSSFP sequence in regards to overall SNR, the arteriovenous contrast and the presence of artifacts.

Imaging parameters for REACT were; Coronal, voxel size=1.2mm³, T2 prep-time=50ms, shot interval=3000ms, TR=3.9ms, TE1/TE2=1.38/2.5ms, flip angle=13°, turbo factor=100 and total acquisition time=2m49s. On the otherr hand, imaging parameters for conventional bSSFP were; Coronal, voxel size=1.2mm³, shot interval=2000ms, TR=4.0ms, TE=2.0ms, flip angle=60°, turbo factor=77, SPIR fat suppression and total acquisition time=3m37s.

In parameter optimization, the CR_Artery-Vein was increased as prep-time increased. Longer prep-time might be better for differentiating the arteries and veins; however, it caused failure water-fat separation in some cases (2/6 cases in prep-time of 70ms and 4/6 cases in 90ms) (Fig.2). Hence, prep-time of 50ms would be optimal within the current sequence.

Figure 3 shows the comparison of REACT and conventional bSSFP sequences. REACT with prep-time of 50ms showed similar CR_Artery-Vein compared to bSSFP. There is no significant differences. On the other hand, CR_{Artery-Muscle} of REACT was significantly higher than that of bSSFP. Moreover, bSSFP caused banding artifacts on the source images in all cases and fat suppression failure in some cases. In contrast, REACT could provide more robust and stable MRA without any artifacts and failed fat suppression. Representative aortic artery images with large FOV are shown in Figure 4.