Measuring cerebrovascular reactivity with breath-hold fMRI in patients with Moyamoya angiopathy: MR perfusion based delay correction significantly improves agreement to [15O]water PET

Abstract

Purpose: Breath-hold functional MRI (bh-fMRI) is able to quantify cerebrovascular reactivity. Vessel stenoses can lead to delayed hemodynamic responses. We aimed to investigate whether delay correction improves the quality of bh-fMRI compared to the diagnostic standard [¹⁵O]water PET.

Methods: The bh-fMRI data sets of 25 patients with Moyamoya Angiopathy were analyzed retrospectively without and with delay correction. Delay correction was calculated using time-to-peak (TTP) maps derived from dynamic susceptibility contrast (DSC) perfusion MRI. [¹⁵O]water PET maps and bh-fMRI maps without and with delay correction were presented blinded for delay correction to two neuroradiologists. The agreement between bh-fMRI without and with delay correction and [¹⁵O]water PET was independently and consensually rated on a 4-point-Likert scale (1 = poor, 2 = moderate, 3 = good, 4 = excellent) and compared with Wilcoxon signed-rank test.

Results: The agreement between bh-fMRI and [¹⁵O]water PET without delay correction was good/excellent (median = 3, modus = 4), and improved significantly after delay correction with medium effect size (median = 4, modus = 4, z = -2.121, p = 0.034, r = 0.42).

Conclusion: Delay correction improves the quality of bh-fMRI and seems to be helpful in clinical practice.

Keywords: Breath-hold fMRI; DCE MRI; Moyamoya angiopathy; [15O]water PET.

Fig. 1

Histogram of the frequency distribution of the intraclass correlation coefficient between the cerebrovascular reactivity without (CVR _uncorrected) and with time delay correction (CVR _corrected) within the individual patients

Fig. 2

Consensus rating of the agreement between the CVR and the [¹⁵O]water PET maps without and with delay correction. Each color represents the data sets of one patient

Fig. 3

Breath-hold fMRI and [¹⁵O]water PET data of one patient who appears to benefit from delay correction. a and b show the mean signal time courses of the right and left middle cerebral artery (MCA-r, MCA-l) and of the cerebellum without (a) and with (b) delay correction. The red vertical lines show the time period of the maximum cerebellar signal change ± 3 s. Without delay correction, the maximum signal changes of the MCA-r and the MCA-l are smaller than the cerebellar maximum signal change but reach values > 0. The delay correction leads to time-shifted mean time courses of the MCA-r and the MCA-l, and results in even negative signal changes (steal phenomenon) at the time period of the maximum cerebellar signal change. The cerebrovascular reactivity (CVR) maps without (c) and with delay correction (d) and the corresponding [¹⁵O]water PET map (e) indicate that delay correction leads to a better agreement between breath-hold fMRI and [¹⁵O]water PET in this patient. The steal phenomenon in the territories of the MCA-r and the MCA-l, as observed in the [¹⁵O]water PET map, are more pronounced in the delay corrected CVR map than in the CVR map without delay correction