CSF Fraction Mapping Derived from Multi‐echo MR FAST‐T2 Sequence Could be an Alternative Measure of Total Perivascular Space in the Brain Parenchyma

Abstract

AbstractBackgroundPerivascular space (PVS), as the key pathway of the glymphatic clearance, has been studied from various aspects. The volume of PVS is considered a biomarker of the enlargement of PVS, which may reflect the deficits of the clearance function. However, most of the methods for the estimation of PVS depends on the segmentation of visible PVS on anatomical MRI, which inherently ignored the invisible PVS that are of sub‐voxel size and dominant of the total PVS. We proposed a muti‐echo T2 relaxometry based CSF water fraction (CSFF) mapping method using 3‐water compartment model. The CSFF mapping corresponding to the long T2 (>200 ms) compartment in the total signal fitted using a 6 echo FAST‐T2 data.MethodForty‐one subjects were recruited (Age: 58.5±17.0, 25 normal (NL), 16 Aβ+ MCI/AD) including 26 females and 15 males. CSFF was reconstructed from the 6 echo FASTT2 data by fitting a three‐water compartment model. PVS in cerebral white matter was segmented using the Frangi filter on the enhanced PVS image. Intracranial volume was segmented using SPM12 for the PVS load normalization.ResultFigure 1 shows the distribution of CSFF and PVS by groups. There is significant difference of CSFF between NL and MCI/AD groups, and no difference of PVS between groups. Figure 2 shows the linear relationship between GM CSFF and normalized PVS load. We see CSFF is about 10 times larger than PVS load. We know the CBV is about 3∼5% of the whole brain. CSFF value is about the same level of CBF, which is very reasonable. It also shows that CSFF has better ability to distinguish NL and MCI/AD than PVS load. Figure 3 shows the high consistency between the CSFF from both high and low resolution FASTT2 data. High resolution CSFF has less partial volume effect, that’s why it has about 0.5% higher value than the low resolution CSFF.ConclusionThe CSF fraction (CSFF) mapping is useful to detect early changes in AD, which may reflect the subtle changes of total PVS space in the brain. Acknowledgement: This study is supported by NIA grant AG057848