Pseudo-continuous and territorial arterial spin labeling MRI for assessment of cerebral perfusion in moyamoya disease after revascularization: A comparative study with digital subtraction angiography

Abstract

Purpose: To evaluate if pseudo-continuous arterial spin labeling (pcASL) and territorial ASL (tASL) can assess cerebral perfusion post-revascularization in Moyamoya disease and compare with digital subtraction angiography (DSA) outcomes.

Materials and methods: Patients diagnosed with Moyamoya disease who underwent pcASL using two post-labeling delays (short ASL, 1,525 ms; delayed ASL, 2,525 ms), tASL, and DSA 3 months after surgery at a single institution were retrospectively evaluated. Manual delineation on pcASL cerebral blood flow (CBF) maps covered middle cerebral artery (MCA) territory on both sides, and cerebellum. Normalized CBF (nCBF) was calculated. Revascularization in the MCA territory was evaluated with external carotid angiography and tASL, graded on a three-point scale. Intermodality agreement was analyzed with weighted κ statistics. Correlation between pcASL-derived nCBF and tASL-measured revascularization, and revascularization grade from direct angiography, was determined. Diagnostic performance of pcASL and tASL was evaluated using DSA as a reference via receiver operating characteristic (ROC) curve analysis.

Results: A total of 32 hemispheres from 31 patients were assessed. On the operated side, sASL and dASL had nCBF values of 1.00 ± 0.30 and 1.31 ± 0.31, respectively. Revascularization area grading showed substantial intermodality agreement (weighted κ = 0.68; 95 % CI: 0.49, 0.87). DSA revascularization moderately correlated with sASL and dASL nCBF values (r = 0.56 and 0.47) and strongly correlated with tASL revascularization area (r = 0.73). ROC analysis revealed that sASL and dASL nCBF values reflected revascularization (area under the curve (AUC) = 0.86 and 0.77) and tASL revascularization area (AUC = 0.91). Combined pcASL and tASL had an AUC of 0.93, comparable to tASL alone, improving diagnostic performance. The diagnostic accuracy of nCBF for sASL was 87.5 %, superior to 75 % for dASL. The diagnostic accuracy of tASL external carotid artery revascularization area was 87.5 %, with sensitivity and specificity of 88 % and 85.7 %, respectively.

Conclusion: The combination of pcASL and tASL outperformed pcASL alone in assessing cerebral perfusion post-Moyamoya disease revascularization.

Keywords: Digital subtraction angiography; Magnetic resonance imaging; Moyamoya disease; Pseudo-continuous arterial spin labeling; Territory arterial spin labeling.

Fig. 1

Images of 42-year-old man diagnosed with right unilateral Moyamoya disease. Patient underwent combined direct and indirect bypass surgery on the right side, as well as digital subtraction angiography (DSA), short arterial spin labeling (sASL), delayed arterial spin labeling (dASL), and territorial arterial spin labeling (tASL) 12 months after surgery. (a, b) Postoperative sASL and dASL showed the same perfusion in the left and right cerebral hemispheres. (c. d) Postoperative lateral (c) and anteroposterior (d) right external carotid arteriograms show areas of revascularization in right MCA territory (more than two-thirds of MCA territory). (e) Right external carotid artery (ECA) territory ASL also depicts supplying territory in right MCA territory (more than two-thirds of MCA territory).

Fig. 2

ROC curve shows sensitivity and specificity values for predicting bypass revascularization outcomes for sASL nCBF, dASL nCBF, and tASL. Predictive value of good prognosis in postoperative tASL (AUC = 0.91, 95 % CI: 0.80–1.00, P < 0.01) was higher than that in sASL nCBF and dASL nCBF. Combined sASL, dASL, and tASL had highest diagnostic performance (AUC = 0.93.95 % CI: 0.81–1.00, P < 0.01).