Purpose

Accelerated MRI using parallel imaging, compressed sensing or a combination of both (Compressed SENSE)¹is widely used in routine clinical examination recently, and it could significantly shorten the total examination time even the contras-enhanced studies. Furthermore, accelerated MRI enables whole-brain 3D acquisition with clinically acceptable scan time. However, there would be a potential risk that such a shorter examination time might miss the appropriate timing for depicting the “delayed-enhanced” lesions, such as brain metastasis and multiple sclerosis plaques^2,3. CENTRA-PLUS⁴ is an optimized centric k-space ordering which can be combined with fat-suppressed turbo field-echo (TFE) and keyhole imaging, it is typically used for abdominal dynamic contrast-enhanced studies to capture early timing after contrast injection. This is also utilized for free-breathing cardiac late-gadolinium enhancement MRI to minimize changes in the contrast kinetics due to washout over a potentially long 3D acquisition time on the order of several minutes⁵.We focused on the CENTRA-PLUS view-ordering framework and tried to modify the k-space profile ordering to enable completely timing-reversed. The central k-space part, most important part for image contrast determination, will be acquired at the almost last of acquisition. We hypothesized that the time-reversed CENTRA-PLUS (SLUP-ARTNEC) can improve the image contrast between delayed-enhanced lesions and background tissues in the contrast-enhanced studies without having the extra-waiting time, resulting in maintain the shorter examination time. The purpose of this study is to demonstrate the feasibility of SULP-ARTNEC in the patients with brain metastasis.

Methods

Theory and pulse sequence:In 3D-TFE sequence, the fat suppression pulse is typically shared over a large number of TFE factors for efficient fat suppression while preventing prolongation of acquisition time. To that end, called low_high radial profile order is typically applied. In this order, from each Ky-Kz segment a profile is acquired per shot, starting with optimal fat suppression at the central Ky-Kz segment. Consequently, the acquisition of the image contrast determining central Ky-Kz profiles are distributed over the entire acquisition. This results in a non-distinct, time-averaged image contrast.
In SULP-ARTNEC, k-space is basically segmented by two (central and peripheral) sectors in the Ky and Kz directions to efficiently obtain the effect of fat suppression pre-pulse. Actual data acquisition is started from the most peripheral k-space sector and then moves in toward centrically. After the complete filling of the peripheral sector, data acquisition is then continued toward central k-space with a centric profile-ordering. Actual k-space acquisition with each method is shown in Figure 1.
As aforementioned, low-high radial order could theoretically be affected by any timing of signals during the acquisition. Whereas, with SULP-ARTNEC, acquisition is started from the peripheral k-space sector and the most central k-space sector is acquired at a late stage. Thus, this may emphasize the increased signals of the brain metastasis due to delayed enhancement at the last of the imaging time (Figure 2).
Experiments: Eight patients with suspected brain metastasis underwent contrast-enhanced MRI on a 3.0T system (Ingenia CX, Philips Healthcare). The study was approved by the local IRB, and written informed consent was obtained from all subjects. We compared two types of 3D T1-TFE with conventional low_high radial order and SULP-ARTNEC. To evaluate the effect of the timing of contrast determination in k-space, each patient underwent three 3D TFE scans after contrast enhancement with either following two acquisition patterns.
Pattern A= Injection (immediately after)-> 1st Conventional-> SULP-ARTNEC-> 2nd Conventional
Pattern B= Injection (immediately after)-> 1st SULP-ARTNEC-> Conventional-> 2nd SULP-ARTNEC
The obtained images were evaluated visually and quantitatively using contrast ration (CR) of neoplastic lesions and cerebral white matter. The scan parameters were as follows: axial, voxel size=0.57x0.57x1.4 mm³ (recon =0.43x0.43x0.7 mm³), FOV=220x220 mm, C-SENSE factor=1.6, SPIR, TR/TE=6.0/2.9ms, and acquisition time=4:40.

Results and Discussion

Representative images of conventional 3D T1-TFE and SULP-ARTNEC without contrast-enhancement in original and reformatted orientations were shown in Fig.3. These images showed similar contrast and image quality.
Fig. 4 shows the example images of pattern A. Bone metastases in the clivus and multiple metastases in the brain were identified. Lesion detection was visually and quantitatively improved over time. SULP-ARTNEC and 2nd Conventional showed close CR because the contrast determination time were close, but 1st Conventional showed lower CR. In addition, there were some brain metastases that enhanced early, but the ring enhancement was more pronounced in the late timing. On the other hand, representative images of pattern B are shown in Fig. 5. A single metastasis in the brain was observed. The CR of the 1st SULP-ARTNEC and Conventional sequences was close, and the CR of the 2nd SULP-ARTNEC sequence was higher. These results suggest that the contrast between metastatic tumors and brain parenchyma was improved over time after contrast enhancement, it is important to take some waiting time. SULP-ARTNEC showed sufficient contrast even it started immediately after the contrast injection, which suggests it is not necessary such an extra waiting time, otherwise, it can further increase the contrast if the waiting time can still be applied.

Conclusion

SLUP-ARTNEC is the unique approach in the post compressed sensing era that can improve the image contrast between delayed-enhanced brain metastasis lesions and background tissues in the contrast-enhanced studies without having the extra-waiting time, resulting in maintain the shorter examination time.

Acknowledgements

No acknowledgement found.