Correlation between regional cerebral blood flow based on simultaneously acquired arterial spin labelling MRI and ¹⁵O-water-PET using zero-echo-time-based attenuation correction

Objectives: Arterial spin labelling (ASL) MRI promises clinical value in several common neurological disorders. Its quantitative accuracy and reproducibility, however, need to be further validated, ideally using simultaneously acquired measurements with ¹⁵O-water-PET on an integrated PET-MR scanner. However, so far, few studies have attempted this and the inclusion of bone in MR-based attenuation correction for PET has thus far been a challenge, compromising the quantitative accuracy of PET-MR based ¹⁵O-water PET data. The aim of the present work was to assess the correlation of ASL- and ¹⁵O-water-PET based regional cerebral blood flow (rCBF) values based on simultaneously acquired data, using zero-echo-time (ZTE)-based attenuation correction, as well as to assess the reproducibility of ASL-based rCBF.

Methods: Six subjects underwent 10 min PET scans after automated bolus injection of 400 MBq ¹⁵O-water (1 mL/s during 5 s followed by 35 mL saline at 2 mL/s) on a time-of-flight integrated PET-MR scanner (Signa PET-MR, GE Healthcare). Arterial blood radioactivity concentrations were monitored using continuous sampling from the radial artery (Swisstrace Twilite Two). Simultaneously, a 3D FSE pseudo-continuous ASL (3D pCASL) with a spiral read-out as supplied by the scanner manufacturer in the commercial software were acquired using an 8 channel head coil (Invivo Hi-Res Head Coil). In addition, 3D T1-w, ZTE and Dixon fat-water MRI were acquired. The ASL procedure was repeated after 2 h (patients remained in the scanner). Quantifiable ASL-based CBF maps were generated. PET images were reconstructed into 26 frames of increasing durations using time-of-flight OSEM (2 iterations, 28 subsets) and a 5 mm post-filter, with ZTE-based attenuation correction. Blood sampler data were corrected for delay and dispersion and ¹⁵O-water-based CBF maps were calculated using a basis function implementation of the single tissue compartment model including a fitted blood volume parameter. CBF maps were co-registered to each patient's T1-w image. 3D T1-w images were segmented and normalised to MNI space using SPM12, and anterior, middle and posterior flow territory volumes of interest (VOIs) were created from a standard template in MNI space and inversely transformed for each patient. In addition, a 45-VOI probabilistic template was applied using PVElab software. Correlations between PET- and ASL-based rCBF values were assessed using regression analysis, and reproducibility of ASL using a paired t-test.

Results: Mean (CI) total brain grey matter CBF values were 67.2 (48.0-86.5) mL/min/100 g for ¹⁵O-water-PET and 65.5 (55.7-75.5) mL/min/100 g for ASL. Although correlation and agreement between ¹⁵O-water and ASL-based rCBF for individual VOIs in the 45-VOI template were generally poor, significant correlations were found on a grey matter flow territory basis, with R² ranging from 0.70 in the anterior flow territory to 0.86 in the middle flow territory. rCBF values were significantly reduced between second and first ASL for all flow territories (p<0.01), with a mean decrease of 10%.

Conclusion: A good correlation between regional flow territory CBF values based on ASL and ¹⁵O-water-PET was found, using ZTE-based attenuation correction for PET data which takes bone tissue into account. ASL values for regional flow territories may have potential applications in patients with dementia or cerebrovascular diseases affecting blood flow such as moya moya. The decrease of ASL-based rCBF values in the reproducibility study needs to be investigated further to assess whether this is a methodological issue or reflects a true decrease in rCBF. Research Support: Uppsala County Council