It is known that multiple sclerosis results in decreased cerebral metabolic rate of oxygen and blood flow^1,2. Both of these are critical elements in cerebrovascular reactivity (CR) to neuronal activation. Using simultaneous BOLD/ASL, we previously measured BOLD activation during a visual task and cerebrovascular reactivity using a hypercapnia challenge in a cohort of multiple sclerosis (MS) patients and matched control subjects³. We showed that the CR is significantly different in MS. That measurement was based on indirect inference that the underlying neuronal activity for a given task is the same and the differences were in CR. Here we further explore this issue by developing a method to directly monitor electrical activity during task performance to understand the relationship between electrical activity, blood flow, and BOLD response.

Data were acquired in 2 subjects, a healthy control (male, age: 51yo) and an MS patient (male, 48yo, EDSS=6). Simultaneous EEG-fMRI was acquired using a 96 channel MRI-compatible EEG system (Brain Products, Germany) and a Siemens 3T Prisma (Erlangen, Germany) with a 32 receive channel head coil.

MR images were acquired using a dual-echo ASL sequence, which provides a perfusion weighted signal from the first echo and a BOLD weighted signal form the second echo⁴. In 7:30 of scan time, and 4 ½ cycles of 32 sec ON/OFF unilateral complex finger tapping were performed. During this time, EEG data were acquired as well.

Data Analysis

EEG Analysis: The EEG data was corrected for scanner artifacts using the MRI master clock signal and volumetric trigger signal, then corrected for ballistocardiogram artifact using an additional ECG signal. Data was decomposed with principal component analysis and the first component selected (linear correlation with appropriate C3/C4 signals was > 0.9, confirming good projection to a cortical loading relevant for the experiment). Component was filtered to various canonical EEG bands (alpha, beta, delta, theta, gamma) using a Morlet transform.

In order to relate the observed electrical activity to BOLD or CBF during task performance, it is necessary to derive a metric from the EEG signal to assess total electrical power during the task compared to rest. This is done by applying a transfer function to translate the observed electrical activity into a timeseries of total power at the temporal resolution of the BOLD measurement^5,6. After testing a number of the transfer functions introduced by Rosa et al.⁶, we determined that the Q_MRF function best modeled the block design BOLD signal.

BOLD-ASL analysis: Briefly, the first 4 volumes from each ASL and BOLD time series are discarded. Perfusion-weighted images are constructed by performing a running subtraction of consecutive control and tagged images using the first acquired echo in each volume. BOLD-weighted images are constructed by performing a running average of the BOLD-weighted second echo. The data were retrospectively motion corrected.Figure 1 shows the simultaneous ASL, BOLD and EEG transfer function for the healthy control during alternating finger tapping and rest. There was a good correspondence in the onset of activation for all signals. The correlation between BOLD and EEG Q_MRF was 0.58 (p<10^-9), between BOLD and ASL was 0.25 (p<0.02), and between ASL and EEG Q_MRF was 0.06 (p<0.5). Figure 2 shows the same timeseries for the MS patient. For the data acquired in the MS patient, correlation between BOLD and EEG Q_MRF was 0.33 (p<0.002), between BOLD and ASL was 0.10 (p<0.36), and between ASL and EEG Q_MRF was 0.10 (p<0.38).In addition to our previous report of impaired CR in MS, at least two other studies report impaired CR in MS, one using ASL techniques examining CBF during tapping and hypercapnia⁷ and the other using resting state BOLD fluctuations⁸. The results shown here indicate that the proposed methods show promise as a way to further understand impaired CR in MS. Further work is necessary to validate the proposed methodology as a way to understand possible differences in CR between healthy control subjects and MS patients and other neurologically compromised populations.