Introduction

Arterial Spin Labeling (ASL) perfusion allows for deeper insights into the physiology of the brain1. Time-resolved information can be used to quantify a number of physiological parameters related to perfusion (e.g. time-to-peak). In combination with artery-selective ASL approaches, collateral flow can be assessed and quantified values can be derived for each feeding vessel individually. Approaches such as Hadamard encoding and super-selective ASL can be used to perform such measurements to obtain time-resolved and vessel-selective information respectively^2,3. However, super-selective ASL usually requires separate measurements for each targeted artery, thus, increase the overall scan time. Here, we present an approach that is based on encoding using a ternary matrix to derive time-resolved information of individually selected major brain-feeding arteries in a single scan.

Materials and Methods

The encoding/decoding process of this study follows a ternary matrix (Fig. 1). In this approach only labeling is performed and no control condition is required. During labeling, the labeling focus of super-selective ASL is moved on top of the selected arteries which have been defined prior to scanning by the operator. This automatic labeling focus displacement during scanning as well as the self-control ASL approach (i.e. no acquisition of control images) were already successfully used earlier⁴. All MR experiments were carried out using a Philips Achieva 3T scanner (Philips Healthcare, Best, The Netherlands). The study was approved by the local ethic committee under the general protocol for MRI pulse sequence development. ASL parameters include: 3600ms total labeling duration and a short post-labeling delay of 50ms. The 3600ms were separated in 9 blocks of 400ms each (Fig. 1). Each acquisition is preceded by a different combination of labeling locations according to a ternary matrix. Within one acquisition, each of the selected arteries is tagged three times during labeling prior to acquisition. Image acquisition was performed as 3D-GraSE scan with 2.3x2.3x5mm3 voxel size covering the whole brain. Scanning the matrix with its 9 acquisitions takes about 32 seconds. The whole procedure is repeated 10 times in order to achieve an adequate SNR. Total scan time is then 5 minutes 24 seconds. Image post-processing was performed according to the post-processing of self-controlled super-selective ASL³. This means the images of the artery of interest are considered twice while the signal of the contralateral arteries adds up and is subtracted from the image of the artery of interest. Thereby an image of a single artery can be calculated⁴.

Results and Discussion

MR experiments show promising results. The images are well delineated despite using no separately acquired control condition. The time-course of perfusion changes can be visualized well (Fig. 2). Having less acquisitions for each artery compared to the Hadamard approach could lead to lower SNR in direct comparison. Therefore, optimizations should be performed to achieve the highest possible SNR using this approach. Furthermore, in the fifth acquisition (2000ms) the left carotid artery is always in tagging condition meaning that this time-point cannot be retrieved by decoding.

Conclusion

Expanding the already in-use Hadamard approach by a third condition allows for the selective visualization of perfusion territories in the human brain without any time delay compared to a conventionally performed ASL scan with the added benefit of obtaining time-resolved and vessel-selective data simultaneously

Acknowledgements

No acknowledgement found.