Extra-neurite Perfusion Measurement with Combined Arterial Spin Labeling and Diffusion Weighted MRI


Extra-neurite Perfusion Measurement with Combined Arterial Spin Labeling and Diffusion Weighted MRI

Asllani, I.; Petr, J.; Mutsaerts, H.-J.; Bozzali, M.; Cercignani, M.

Introduction:

Arterial Spin Labeling (ASL) is an MRI method that uses magnetically labeled endogenous water as a tracer for measuring cerebral perfusion in vivo1. The arterial water that is usually 'labeled' at a plane positioned at the base of the brain, perpendicular to the carotids. A post-labeling delay (PLD) is introduced prior to acquisition to allow labeled water to cross the vasculature and perfuse into the tissue1. Because of signal decay due to T1 relaxation, fast acquisition schemes are employed to ensure optimal SNR. Consequently, the spatial resolution of ASL is relatively low (~ 3 x 3 x 6 mm3). As such, the measured blood flow from a given voxel reflects a mixture of signals from gray matter (GM), white matter (WM), and CSF, a phenomenon known as partial voluming (PV)2. To correct for the confounding effects of PV in ASL imaging, an algorithm (PVC) has been developed and already used by several studies2,3. The algorithm is based on GM and WM volume data obtained from the segmentation of the T1w image2, and makes no further distinction between different compartments within the same tissue type. Here, we investigated the potential of PVC ASL to map blood perfusion in the extra-neurite compartment (e.g., soma, glial cells4) and the intra-neurite (comprised of axons and axon terminals4) within the same tissue, independently. We applied the PVC algorithm using compartmental data from a diffusion weighted imaging (DWI) model, referred to as NODDI4. The underlying hypothesis was that the blood flow in the extra- and intra-neurite compartments would vary with the PLD; a short PLD acquisition would increase the flow in the extra-neurite compartment compared to the long PLD for which there should be an increased flow in the intra-neurite compartment instead.
Methods:
Theory
At any given voxel, the blood flow (fT) is given as:
fT=VFIn•fIn+VFEn•fEn+VFIso•fIs
where, VFIn, VFEn, VFIso represent respectively: the intra-neurite, extra-neurite, and non-tissue compartments obtained from NODDI4. By assuming that for each compartment blood flow is constant over a 'kernel', the equation can be re-written in vectorial form to reflect the flow at the voxel in the center of the kernel2, from which then each compartmental flow can be computed using linear regression as detailed in Asllani et al.2..

MRI protocol & image analysis
T1w (MPRAGE), NODDI, and ASL MRI images were obtained on 4 healthy participants (mean age = 44.5 ± 7.4 y, 2 men) a Siemens 3T system. To test the hypothesis that a shorter PLD would increase the signal in the extra-neurite GM compartment, ASL was acquired with a short (200ms) and long PLD (1800ms). Only results from voxels with GM content > 80% are presented.
Results:
Fig.1 shows the raw images that were used by the PVC algorithm to extract the flow from each compartment within the GM. For the long-PLD acquisition, average CBF in the extra- and intra-neurite compartments was 76 ± 10 mL/100g*min and 59 ± 8 mL/100g*min, respectively. As hypothesized, for the short-PLD, the CBF signal was contained primarily in the extra-neurite department (118 ± 17 mL/100g*min) with the intra-neurite compartment flow being essentially zero (-0.9 ± 0.6 mL/100g*min). Results from one participant are shown in Fig.2.
Supporting Image: Fig1.jpg
·Fig.1: ‘Raw’ NODDI and ASL images used by the PVC algorithm from one subject. Top row: MPRAGE and VFIn images; middle row: VFEn and VFISO; bottom row: CBF for short PLD (left) and long PLD (right).
Supporting Image: Fig2.jpg
·Fig.2: Top: Extra-neurite GM CBF from short (left) & long (right) PLD acquisitions. Bottom: axial and sagittal views of Intra-neurite CBF for long PLD with areas in blue indicating ~zero signal.

Conclusions:
We combined NODDI with PVC ASL MRI to distinguish between blood flow in the extra- and intra-neurite compartments within GM. While these initial results look promising, more work is needed to test the sensitivity of this method and its feasibility for clinical applications. For example, a larger PLD range is needed to test whether the method can be used to detect inter-neurite subcortical flow. If successful, this method could prove invaluable in mapping blood flow with high spatial specificity.

Keywords: Cerebral Blood Flow; Data analysis; fMRI CONTRAST MECHANISMS; MRI

  • Open Access Logo Contribution to proceedings
    Organization for Human Brain Mapping Annual Meeting 2019, 09.-13.06.2019, Rome, Italy
  • Open Access Logo Poster
    Organization for Human Brain Mapping Annual Meeting 2019, 13.06.2019, Rome, Italy

Downloads

Permalink: https://www.hzdr.de/publications/Publ-29417
Publ.-Id: 29417