Partial volume correction for quantitative CEST imaging of acute ischemic stroke

Abstract

Purpose Contributions of cerebrospinal fluid (CSF) have not been previously taken into account in the quantification of APT CEST effects, and correction for the dilution of CEST effects by CSF may allow for more robust measurement of CEST signals. The objective of this study was to compare the robustness of a partial volume (PV) correction model against a standard (4‐pool) multi‐pool model as far as their ability to quantify CEST effects in healthy, normal, and pathological tissue. Methods MRI data from 12 patients presenting with ischemic stroke, and 6 healthy subjects, were retrospectively analyzed. CEST signals derived from a 4‐pool model and a PV correction model were compared for repeatability and pathological tissue contrast. The effect of PV correction (PVC) was assessed within 3 ranges of tissue PV estimate (PVE): high PVE voxels, low PVE voxels, and the whole slice. Results In voxels with a high tissue PVE, PV correction did not make a significant difference to absolute APTR*. In low PVE voxels, the PVC model exhibited a significantly decreased ischemic core signal. The PVC measures exhibited higher repeatability between healthy subjects (4 pools: 3.4%, PVC: 2.4%) while maintaining a similar ischemic core CNR (0.7) to the 4‐pool model. In whole slice analysis it was found that both models exhibited similar results. Conclusions PV correction yielded a measure of APT effects that was more repeatable than standard 4‐pool analysis while achieving a similar CNR in pathological tissue, suggesting that PV‐corrected analysis was more robust at low values of tissue PVE.

Clinical translation of amide proton transfer (APT) MRI for ischemic stroke: a systematic review (2003–2020)

“…The number of published clinical studies showed a gradual increase over the years. In the 14 studies, a total of 282 patients had been scanned with APT imaging after accounting for those judged to be duplicate analyses (using the same patient data but for different purposes); the study by Harston et al (33) and two studies by Msayib et al (34,35) (26,(33)(34)(35)(36)(37), three studies involved only acute stroke patients (29,38,39), one study included both hyperacute and acute patients (40), two studies involved acute and early subacute patients (41,42), while two studies involved patients from the hyperacute to early subacute stages (30,43). Further information on these studies such as the treatment received by patients, scanner field strength, radio frequency (RF) saturation scheme, imaging acquisition parameters, image acquisition time, APT effect quantification method, and other MRI sequences acquired, are presented in https://cdn.amegroups.cn/static/public/qims-20-1339-1.…”

“…Six of the included studies involved only hyperacute patients, scanned within 24 hours from onset (26,(33)(34)(35)(36)(37), whereas the other eight studies involved patients that were scanned outside of this time frame (29,30,(38)(39)(40)(41)(42)(43). Among which, three studies investigated the progression of the disease with time, namely the studies by Song et al (40), Yu et al (30), and Momosaka et al (43).…”

Enhanced CEST MRI Using the Residual of Inversed Z-Spectra for Ischemia Detection

“…Recently, there have been several efforts to introduce other approaches for analyzing CEST data, including a three-offsets method [9], extrapolated semi-solid MT model reference approach [10], multiple-pool Lorentzian fitting [11,12] and Lorentzian difference (LD) analysis [13]. However, three-offsets method assumes a simplistic linear baseline, which is limited to high-field application [14]. EMR can achieve higher detection sensitive and allows a reliable delineation of ischemic penumbra tissues [15], however, this method requires longer scanning time.…”

Quantitative chemical exchange saturation transfer imaging of nuclear overhauser effects in acute ischemic stroke

“…The amide pool prior mean was 90 mM, with a standard deviation (SD) of 20 mM for the human data and 200 mM for the preclinical data. The multipool model‐based analysis is implemented in Quantiphyse, made available for download at https://www.quantiphyse.org 15,19,37 …”

“…The multipool model-based analysis is implemented in Quantiphyse, made available for download at https:// www.quantiphyse.org. 15,19,37 In previous studies MT effects arising from semisolid tissue have been quantified using the model-free conventional MT ratio (MTR) metric, defined as MTR = S 0 −S(50 ppm) S 0…”