Edited by: Olga Boukrina, Kessler Foundation, United States
Reviewed by: Harvey Pollard, Uniformed Services University of the Health Sciences, United States; Antonino F. Germano’, University of Messina, Italy
Specialty section: This article was submitted to Neurotrauma, a section of the journal Frontiers in Neurology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Blood oxygenation level-dependent (BOLD) functional MRI (fMRI) has been extensively used as a marker of brain dysfunction and subsequent recovery following stroke. However, growing evidence suggests that straightforward interpretation of BOLD fMRI changes with aging and disease is challenging. In this study, we investigated the effect of calibrating task fMRI data by applying a hemodynamic calibration method using the resting-state fluctuation amplitude (RSFA). Task fMRI responses were obtained during a covert verbal fluency task in a group of early stage stroke patients and matched healthy normal controls.
Fifteen acute left hemisphere stroke patients (less than 7 days from stroke; aged 44–84 years, average ~64 years) and 21 healthy controls (aged 55–77 years, average ~61 years) were prospectively studied. All subjects completed a 3-min covert verbal fluency task, and a 10-min eyes-closed resting-state fMRI scan, from which the calibration factor (RSFA) was computed. A behavioral measure on the verbal fluency task was also collected outside the scanner. Whole brain activation volumes and region-of-interest (ROI)-wise percent signal change and activation volumes before and after calibration were computed.
Between-group differences in whole brain activation volumes, although statistically significant before calibration failed to be significant after calibration. There were significant within-group differences before and after calibration with RSFA. Statistically significant between-group differences on ROI-wise measures before calibration also significantly reduced after calibration. Exploratory brain-behavior correlations revealed a similar pattern: significant correlations before calibration failed to survive after calibration.
BOLD fMRI changes with aging and disease is confounded by changes in neurofunctional coupling leading to challenges in the straightforward interpretation of task fMRI results. Application of the hemodynamic calibration using the RSFA technique in the current study appeared to mitigate any differences between stroke and age-matched healthy controls. Our study indicates that estimating neural activity after applying hemodynamic scaling is important for studies of aging and for studies tracking post-stroke changes. We recommend that further investigation of hemodynamic calibration with RSFA in healthy subjects and in stroke in larger samples is necessary.
The blood oxygenation level-dependent (BOLD) contrast in functional MRI (fMRI) studies has been extensively used to study group differences and to relate brain response to overt behavior. The BOLD signal is a function of changes in cerebral blood flow (CBF), volume, and oxygenation. A key challenge, however, in investigating BOLD fMRI task-associated responses in aging and disease is that there may be considerable variability in the responses across subjects depending on the intrinsic vascular reactivity to neural activity (
Several studies have used a breath-hold task in the MR scanner to measure the brain’s cerebrovascular response to a stimulus and a hypercapnic hemodynamic calibration method to eliminate inter-subject variability in task fMRI studies [e.g., Ref. (
The current study is part of an ongoing research study investigating brain reorganization changes following stroke. Data from 15 early left hemisphere stroke patients (mean age = 64 years, range 44–84 years, 11 males, 12 right handed) and 21 healthy controls (mean age = 62.7 years, range 55–77 years, 11 males, 20 right handed) were analyzed for this study.
The demographic characteristics of all participants are shown in Table
Demographic characteristics of healthy controls in the study.
Healthy control | Age | M/F | Handedness | Education (years) | Acute stroke | Age | M/F | Handedness | Education (years) |
---|---|---|---|---|---|---|---|---|---|
1 | 71 | F | R | 18 | 1 | 67 | M | L | 21 |
2 | 67 | M | R | 21 | 2 | 58 | M | R | 16 |
3 | 74 | F | R | 16 | 3 | 79 | M | R | 21 |
4 | 61 | F | R | 12 | 4 | 58 | F | R | 12 |
5 | 55 | M | R | 21 | 5 | 70 | M | R | 20 |
6 | 58 | M | L | 18 | 6 | 63 | M | L | 12 |
7 | 60 | F | R | 16 | 7 | 55 | M | R | 12 |
8 | 60 | M | R | 17 | 8 | 62 | M | R | 16 |
9 | 55 | F | R | 16 | 9 | 75 | M | R | 18 |
10 | 59 | F | R | 21 | 10 | 84 | M | L | 16 |
11 | 63 | F | R | 18 | 11 | 44 | F | R | 16 |
12 | 62 | F | R | 18 | 12 | 59 | F | R | 18 |
13 | 56 | M | R | 22 | 13 | 62 | F | R | 16 |
14 | 61 | M | R | 14 | 14 | 62 | M | R | 14 |
15 | 62 | F | R | 18 | 15 | 61 | M | R | 14 |
16 | 77 | M | R | 14 | |||||
17 | 62 | M | R | 16 | |||||
18 | 62 | F | R | 18 | |||||
19 | 63 | M | R | 16 | |||||
20 | 61 | M | R | 16 | |||||
21 | 68 | M | R | 18 |
Clinical characteristics of early stroke patients.
Patient | Age | Time since stroke (days) | NIH stroke score | Lesion location |
---|---|---|---|---|
1 | 67 | 5 | 0 | Left lateral medulla |
2 | 58 | 5 | 0 | Left corticospinal tract |
3 | 79 | 6 | 2 | Left MCA |
4 | 58 | 7 | 1 | Left frontal |
5 | 70 | 2 | 2 | Left parietal |
6 | 63 | 2 | 2 | Left putamen |
7 | 55 | 3 | 0 | Left MCA |
8 | 62 | 9 | 0 | Left parietal |
9 | 75 | 6 | 1 | Left occipital |
10 | 84 | 5 | 1 | Left MCA |
11 | 44 | 5 | 7 | Left insula, L frontal |
12 | 59 | 2 | 2 | Left posterior insula, left angular gyrus |
13 | 62 | 5 | 2 | Left MCA |
14 | 62 | 12 | 0 | Left paramedian midbrain |
15 | 61 | 5 | 0 | Left peri thalamic region and left corona radiate |
Verbal fluency outside the scanner was assessed by forms of the Controlled Oral Word Association Test (
A 3-T GE whole-body MRI scanner (GE Healthcare, Waukesha, WI, USA) equipped with a 8-channel head coil was used to acquire the MRI images. An axial localizer scan was obtained to verify subject positioning and plan slice acquisition. T1-weighted axial anatomical images were acquired at the beginning of each session using FSPGR BRAVO sequence (TR = 8.132 ms, TE = 3.18 ms, TI = 450 ms, 256 × 256 matrix, 156 slices, flip angle = 12°, FOV = 25.6 cm, slice thickness = 1 mm). Ten minutes, eyes closed, resting-state scans were obtained using single-shot echo-planar T2*-weighted imaging with the following acquisition parameters: TR = 2.6 s, 231 time-points, TE = 22 ms, FOV = 22.4 cm, flip angle = 60°, voxel dimensions 3.5 mm × 3.5 mm, 3.5 mm slice thickness, 40 slices. Functional data were acquired
All preprocessing of task and resting fMRI data was performed using the AFNI package (
All preprocessing was done in subject space and then images normalized to MNI space for computing uncalibrated and calibrated BOLD signal amplitude and activation volumes.
In order to evaluate the RSFA calibration method at the ROI level, an almost identical process as outlined above was used for computing the BOLD percent signal change and activation volumes with the key difference that the 99th percentile threshold value was computed for each ROI to compute BOLD signal amplitude and activation volumes at an uncorrected significance threshold of
Chi-squared tests were run to examine group differences on demographic variables of gender and handedness. Independent
Chi-squared tests showed no significant differences between the groups on gender (
There were significant differences between stroke patients and healthy controls on the letter fluency task raw scores (mean = 27 in acute strokes vs. 51 in healthy controls,
Axial view of the group activation maps before calibration during the verbal fluency task in controls (top), patients (middle), and difference map (controls > patients, bottom panel). Maps are in radiological convention, left = right,
Bartlett’s test showed no differences in variance in whole brain activation volume between groups (
Prior to calibration with RSFA, whole brain group activation volumes were significantly smaller in the patient group, compared to the controls during the verbal fluency task (
Axial view of the group activation maps after calibration with RSFA during the verbal fluency task in controls (top), patients (bottom panel). Maps are in radiological convention, left = right,
Regions of activation (uncalibrated) in controls and patients on the verbal fluency task.
Montreal Neurological Institute coordinates |
||||||
---|---|---|---|---|---|---|
Controls | Region | Voxels | Max intensity |
|||
1 | +6 | −18 | +48 | L supplementary motor area (SMA) | 1,779 | 7.48 |
2 | −33 | 60 | −30 | R cerebellum | 994 | 7.91 |
3 | −30 | −21 | 12 | R insula | 464 | 7.28 |
4 | −63 | 48 | 18 | R superior temporal gyrus | 383 | −6.34 |
5 | 33 | 42 | 42 | L inferior parietal lobule | 356 | 6.68 |
1 | −60 | +54 | +24 | R angular gyrus | 149 | −6.33 |
2 | −27 | +57 | −21 | R cerebellum | 114 | 7.91 |
3 | +45 | +9 | +54 | L postcentral gyrus | 87 | 5.84 |
4 | +45 | −33 | +24 | L inferior frontal gyrus (p. triangularis) | 50 | 6.36 |
5 | +3 | −3 | +66 | L SMA | 43 | 6.95 |
1 | 39 | −9 | +24 | L inferior frontal gyrus (p. opercularis) | 144 | 7.45 |
2 | −51 | +54 | −12 | R inferior temporal gyrus | 96 | 6.44 |
3 | −24 | +72 | −54 | R cerebellum | 82 | 7.69 |
4 | +39 | +66 | −3 | L inferior occipital gyrus | 63 | 4.51 |
5 | −24 | +69 | −27 | R cerebellum | 46 | 7.91 |
After calibration with RSFA, whole brain group activation volumes remained smaller in the patient group, compared to the controls during the verbal fluency task, but this difference was not statistically significant (
Regions of activation (calibrated) in controls and patients on the verbal fluency task.
After calibration |
Montreal Neurological Institute coordinates |
|||||
---|---|---|---|---|---|---|
Controls | Region | Voxels | Max intensity |
|||
1 | 30 | −24 | 3 | L insula | 853 | 5.16 |
2 | −60 | 45 | 27 | R supramarginal gyrus | 513 | 5.16 |
3 | 45 | 66 | −12 | L inferior occipital gyrus | 147 | 4.26 |
4 | −39 | −54 | −3 | R middle frontal gyrus | 123 | 4.26 |
5 | −24 | 9 | −24 | R parahippocampal gyrus | 104 | 4.26 |
1 | 33 | −9 | −9 | L insula | 75 | 5.83 |
2 | 45 | 0 | 9 | L rolandic operculum | 67 | 4.12 |
3 | 42 | 63 | −15 | L fusiform gyrus | 47 | 4.12 |
4 | −30 | −18 | −12 | R insula | 44 | 4.61 |
5 | 6 | 45 | −63 | L cerebellum | 43 | 4.12 |
Within- and between-group differences in whole brain activation volumes.
Activation volumes reduced in both groups after calibration. Between-group differences were significant prior to calibration in two of four regions but these differences were not significant after calibration. Within-group differences from pre- to post-calibration were significant for three of four regions in controls, and for one region, the left IFG1, in patients (Table
Region-of-interest-wise group activation volumes (number of voxels) and
Left inferior frontal 1 |
Left inferior frontal 2 |
Right inferior frontal 1 |
Right inferior frontal 2 |
|||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Uncalibrated | Calibrated | Uncalibrated | Calibrated | Uncalibrated | Calibrated | Uncalibrated | Calibrated | |||||
Controls | 106 | 36 | 105 | 35 | 73 | 33 | 62 | 31 | 0.05 | |||
Patients | 65 | 32 | 61 | 28 | 0.026 | 52 | 31 | 0.120 | 46 | 28 | 0.15 | |
0.204 | 0.322 |
Region-of-interest (ROI)-wise activation volumes—older controls show statistically significant difference before and after calibration in all regions except the right IFG2 (
Blood oxygenation level-dependent amplitude increased post-calibration in controls but not in patients. Between-group differences were, however, significant only prior to calibration in two of four regions; there were no significant differences after calibration in any of the ROIs. Within-group differences from pre- to post-calibration were significant for three of four regions, only in controls (Table
Region-of-interest-wise percent signal change and
Left inferior frontal 1 |
Left inferior frontal 2 |
Right inferior frontal 1 |
Right inferior frontal 2 |
|||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Uncalibrated | Calibrated | Uncalibrated | Calibrated | Uncalibrated | Calibrated | Uncalibrated | Calibrated | |||||
Controls | 0.53 | 1.04 | 0.93 | 1.03 | 0.357 | 0.72 | 1.02 | 0.64 | 1.03 | |||
Patients | 1.01 | 1.00 | 0.98 | 1.63 | 0.96 | 0.021 | 1.21 | 0.99 | 0.385 | 1.22 | 1.00 | 0.489 |
0.687 | 0.494 | 0.034 | 0.755 | 0.037 | 0.820 |
Region-of-interest (ROI)-wise blood oxygenation level-dependent amplitude defined as conventional percent signal change for uncalibrated data and tSD (task) ÷ tSD (rest) (temporal SD of task divided by temporal SD of rest) for calibrated data: older controls show statistically significant difference before and after calibration in all regions except the left IFG2 (
Exploratory Pearson correlations between ROI-wise brain measures and verbal fluency normed scores showed trend toward significance only in patients. For activation volume, prior to calibration, Pearson
Task fMRI has been extensively used to map between-subject variability in healthy controls and to track post-stroke recovery (
The main objective of this work was to investigate the use of a calibration parameter in the analysis of the effects of stroke on task fMRI BOLD activation. We compared results from stroke patients with a group of age-matched healthy controls before and after calibration. The calibration measure we investigated, “resting-state fluctuation amplitude (RSFA),” derived from a 10-min resting fMRI scan, may be a useful proxy for cerebrovascular reactivity, especially in patient populations, where the patient’s ability to perform a breath-hold task could confound results.
Based on task fMRI studies of language recovery following stroke, it has been proposed that in the early stages there is little activation in the left hemisphere followed by a more bi-hemispheric activation in the sub-acute stage and subsequent restoration of activation in the left hemisphere in the chronic stages of stroke (
In the current study, there were significant group differences in behavioral performance on the verbal fluency task that is consistent with prior studies (
Similar to the whole brain results, ROI-wise comparisons also showed that significant between-group differences in activation volumes or the BOLD amplitude in the left inferior frontal regions disappeared after calibration. Interestingly, although activation volumes continued to be slightly greater in controls than patients, ROI-wise amplitudes after calibration appeared to be similar in patients and controls. These mixed results therefore indicate that further investigation into these differences with other hemodynamic calibration techniques is warranted (
A basic assumption in fMRI studies investigating group differences is that the two groups have comparable neurovascular coupling. However, in the face of pathology that affect this coupling in one group (e.g., stroke) but not the other (or perhaps a different pathology in the other, e.g., aging) make these comparisons extremely difficult. So when calibration mitigates group differences, as reported in this study, it is difficult to determine if a previously existing difference (in the uncalibrated state) is now obscured by calibration, or calibration merely identified additional regions of activation for both groups, thereby reducing group differences in activation. There may be other factors that may capture the clinical deficit such as functional and structural network connectivity differences between the group, if activation differences do not capture it.
Not only alterations in the cerebral vasculature but also alterations in the complex neurochemical transformation of neural activity to changes in blood flow could also affect the measured BOLD response (
Our study has several limitations, including relatively small sample size and studying both cortical and subcortical stroke patients. Additionally, our study differs from previous analyses using RSFA in that we have taken extra steps to reduce motion artifact, which may have confounded previous reports. Nonetheless, we found mixed results when scaling with RSFA. RSFA itself as a calibration factor is relatively novel but is known to be sensitive to neural activity and head motion in addition to physiological changes such as variations in breathing (
Because resting fMRI can be easily acquired during a MR scan session, we recommend that hemodynamic calibration of BOLD fMRI activation in stroke patients with the RSFA should be further explored and the nature of the RSFA should be further explored as well since it may potentially be a convenient calibration factor. Although studies with larger sample sizes are needed to evaluate the effect of such calibration on brain–behavior relationships, our current study provides a first step in the application of hemodynamic calibration using the RSFA technique.
This study was carried out in accordance with the recommendations of the University of Wisconsin-Madison Health Sciences Institutional Review Board, with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the UW-Madison IRB.
VN was involved in recruitment, data collection, data analysis, and writing. RR was involved in data analysis and writing. VP is the lead PI on the study and supervised all aspects of this work.
The authors have no conflicts of interest to report, as this research was conducted in the absence of commercial and financial relationships that might compromise the integrity of the results reported herein.