Sage Journals: Discover world-class research

Abstract

We used voxel-based morphometry (VBM) to compare grey matter volume (GMV) between 20 migraine patients (five with aura and 15 without aura) with normal conventional magnetic resonance imaging findings and 33 healthy controls matched for age and sex. A separate analysis was also performed to delineate a possible correlation between the GMV changes and the headache duration or lifetime headache frequency. When compared with controls, migraine patients had significant GMV reductions in the bilateral insula, motor/premotor, prefrontal, cingulate cortex, right posterior parietal cortex, and orbitofrontal cortex (P < 0.001, uncorrected for multiple comparisons at a voxel level; corrected P < 0.05 after small volume corrections). All regions of the GMV changes were negatively correlated with headache duration and lifetime headache frequency (P < 0.05, Pearson's correlation test). We found evidence for structural grey matter changes in patients with migraine. Our findings of progressive GMV reductions in relation to increasing headache duration and increasing headache frequency suggest that repeated migraine attacks over time result in selective damage to several brain regions involved in central pain processing.

Keywords

Grey matter migraine voxel-based morphometry

Introduction

Migraine is a form of primary neurovascular headache that is likely to be based on dysfunction in the brain, and functional neuroimaging studies have contributed greatly to the understanding of the pathophysiology of migraine (1, 2). Specifically, positron emission tomography (PET) studies have shown neuronal activation during migraine attacks in several brain regions that are known to be generally involved in pain processing (i.e. insula, cingulate cortex, prefrontal cortex and thalamus) (3–7). These studies have also demonstrated activation in the brainstem, particularly the dorsolateral pons, which is now recognized as a specific region involved in the generation of migraine headache (3–7).

Apart from the considerable number of functional neuroimaging studies regarding acute migraine attacks, only few studies have examined the structural changes associated with migraine (8–10). As a form of primary headache, it is generally accepted that there are no macroscopic structural changes in migraine (2, 11). Voxel-based morphometry (VBM) is a fully automated, unbiased, operator-independent magnetic resonance imaging (MRI) analysis technique that is being increasingly used to detect subtle structural differences in brain tissue composition on a voxel-wise comparison between groups of subjects (12). A recent study using VBM found significant grey matter (GM) abnormalities in several brain regions known to be involved in central pain processing in migraine patients with evidence of T2-visible white matter (WM) lesions (10). Although subclinical T2-visible WM lesions are not infrequently seen in migraine patients, the most patients have normal MRI findings (13). Given that both migraine with and without WM lesions share a common pathophysiological mechanism, the question arises whether such GM changes in migraine patients with WM lesions could be observed in migraine patients without WM lesions. In the present study, we used VBM to replicate the previously published findings (10) in migraine patients without WM lesions in an attempt to generalize these findings to the entire migraine population. We also searched for possible correlations between changes in GM volume (GMV) and headache duration or lifetime headache frequency.

Methods

Subjects

Consecutive patients with episodic migraine, based on the criteria established by the International Headache Society (11, 14), were recruited from the headache population who visited the out-patient clinic in the Department of Neurology at Korea University Medical Centre from July 2006 to November 2006. Patients with hypertension, diabetes mellitus, hypercholesterolaemia, heart disease, other chronic systemic diseases, stroke, cognitive impairment, substance abuse, as well as other neurological or psychiatric disorders were excluded. Among 22 migraine patients who underwent MRI examination, two patients (female; age 38 and 50 years) with any WM lesions on T2-weighted MRI were excluded. The remaining 20 right-handed patients with episodic migraine (five with aura, 15 without aura; three male; age range 15–53 years; mean age 33.7 ± 11.3 years) were selected for the study. Mean age at migraine onset was 23.9 ± 7.4 years (range 9–35 years), mean duration of disease was 9.8 ± 6.0 years (range 2–23 years), mean duration of migraine attacks was 21.8 ± 13.1 h (range 6–48 h), mean headache pain intensity measured by a 4-point scale (0, none; 1, mild; 2, moderate; 3, severe) was 2.5 ± 0.6 (range 1–3), and mean frequency of migraine attacks per year was 32.7 ± 10.9 (range 12–60). Estimated lifetime frequency of migraine attacks (average frequency of migraine attacks per year × duration of the disease in years) ranged from 36 to 864 (mean 347.1 ± 257.7). All patients had been free from a typical migraine attack for at least 1 week prior to MRI examination. Abortive medications that the patients usually took for their migraine attacks included oral triptans in 13 patients, ergotamine-caffeine preparations in five, and non-steroidal anti-inflammatory drugs in two. None of the patients was on any prophylactic medications for migraine at the time of MRI examination. Thirty-three age- and sex-matched, right-handed healthy volunteers (four male; age range 15–53 years; mean age 33.8 ± 10.5 years; P = 0.97 vs. patients) with no family history of migraine, extremely few (less than a few days per year) spontaneous non-throbbing headaches, no history of substance abuse, and no history of systemic, psychiatric or neurological disorders served as controls. This study was approved by the Ethics Committee of Korea University Medical Centre, and written informed consent was obtained from all patients and controls.

MRI acquisition and voxel-based morphometry

MRI examination was performed on a 1.5-T scanner (Siemens Sonata, Erlangen, Germany) in all patients and controls. Whole-brain 3-D T1-weighted gradient echo images were acquired for each subject using the magnetization-prepared rapid-acquired gradient echoes (MP-RAGE) sequence (TR, 2200 ms; TE, 3.7 ms; TI, 1100 ms; NEX, 1.0; field-of-view 24 × 24 cm; matrix size 256 × 256; slice thickness 1.25 mm), yielding 128 contiguous sagittal slices with a defined voxel size of 0.94 × 0.94 × 1.25 mm. Together with the volumetric data, conventional MR images were also obtained by using the following sequences: T1-weighted axial, T2-weighted axial/coronal, fluid-attenuated inversion recovery (FLAIR) axial, and gadolinium-enhanced T1-weighted axial/coronal images (5 mm thickness for each sequence). Two experienced neuroradiologists (S.-I.S. and H.Y.S.), blinded to each subject's clinical diagnosis, independently reviewed the conventional MR images of all patients and controls.

Optimized VBM protocol (15) was employed for data processing and analysis, with SPM2 (Wellcome Department of Cognitive Neurology, London, UK) implemented in Matlab 7.0 (Mathworks, Natick, MA, USA). Briefly, in order to yield optimal spatial normalization and segmentation of the images, a study-specific template of the whole brain and tissue probability maps (priors) were created from all subjects participating. Initially, all images were spatially normalized to the standard MNI T1 MRI template (ICBM 152 standard) using linear 12-parameter affine transformation, segmented into GM, WM and cerebrospinal fluid (CSF) compartments using MNI priors, smoothed with an 8-mm full width at half maximum (FWHM) isotrophic Gaussian kernel, and averaged to obtain study-specific priors. The segmentation method also included an automated brain extraction procedure to remove any non-brain tissue and an algorithm to correct for image intensity non-uniformity. Next, the original structural images were segmented into GM, WM and CSF in native space using the study-specific template and priors. The extracted GM and WM images were then normalized to the study-specific GM and WM templates using combined 12-paramenter linear and non-linear (7 × 8 × 7 basis functions) transformations. The parameters derived from this spatial normalization step were then reapplied to the original structural images. These fully normalized images were resliced to a final voxel size of 1 mm³, and then segmented into GM, WM and CSF compartments using the study-specific priors. The segmented images were modulated by the Jacobian determinants derived from the spatial normalization step in order to compensate for the possible volume changes during the non-linear spatial normalization procedure. Finally, these normalized, segmented and modulated GM images were smoothed using a 10-mm FWHM kernel. The analysis of modulated GM images can detect regional differences in GMV.

Statistical analysis

The normalized, modulated and smoothed GM data were analysed using SPM2, within the framework of General Linear Model on a voxel-wise comparison. Regionally specific differences in GMV between migraine patients and controls were assessed using an analysis of covariance (ancova) with total intracranial volume, age and sex treated as nuisance variables. An absolute GM threshold of 0.2 (of a maximum value of 1) was used to avoid possible edge effects around the border between GM and WM. Two contrasts were defined to examine both increases and reductions in GMV between the groups. Statistical parametric maps were thresholded at P < 0.001, uncorrected for multiple comparisons at a voxel level across the whole brain, and an extent threshold of 200 contiguous voxels was applied. Given our a priori knowledge of the involvement of brain regions related to migraine or acute pain (10, 16), we further applied a small volume correction (SVC) to these regions using a 12-mm radius with a threshold of P < 0.05, corrected for multiple comparisons. The anatomical localization of each peak cluster was established based on their coordinates by using Talairach Daemon software (17). Separate correlation analyses were performed in order to delineate a possible relationship between the identified GMV changes and the headache duration and between the GMV changes and the lifetime headache frequency. For this purpose, we extracted the normalized voxel values of each significant cluster using the ‘volume-of-interest (VOI)’ function of SPM2, and correlated these values with the headache duration in years and estimated lifetime frequency of migraine attacks (Pearson correlation test, P < 0.05; SPSS Version 12.0; SPSS Inc., Chicago, IL, USA).

Results

The conventional MR images of all patients and controls were reported as normal: neither unusual nor abnormal findings, including T2-visible WM lesions, could be found. No significant between-group differences were found in global brain volume (calculated as the sum of GM, WM and CSF volumes; patients, 1487 ± 127 ml; controls, 1502 ± 116 ml; P = 0.67) or total GMV (patients, 717 ± 61.2 ml; controls, 720 ± 58.4 ml; P = 0.86). When compared with controls, migraine patients had significant GMV reductions in the following regions: bilateral insula (BA 13), bilateral motor/premotor cortex (BA 4/6/8), bilateral prefrontal cortex (BA 9/10/44/46), left dorsal anterior cingulate cortex (BA 32), right dorsal posterior cingulate cortex (BA 31), right inferior (BA 40) and superior parietal cortex (BA 7), orbitofrontal cortex (BA 11) and visual cortex (BA 18) (Fig. 1, Table 1). All of these regions, except for the visual cortex (BA 18) where the a priori hypothesis is not available, remained significant after SVCs (P < 0.05, corrected for multiple comparisons; Table 1). No regions of significant GMV increases were found at the same threshold. Correlation analyses showed that all regions of the GMV changes were significantly negatively correlated with the headache duration and estimated lifetime frequency of migraine attacks (P < 0.05, Pearson correlation test; Table 1).

Table 1

Regions of significant reductions in grey matter volume in migraine patients compared with controls

Location	BA	Coordinates (mm)			Peak Z	Corrected P ^SVC	r (P-value)∗	r (P-value)†
x	y	z
L insula	13	−41	−5	−2	4.89	0.000	−0.72 (0.000)	−0.67 (0.001)
R insula	13	42	−15	1	4.45	0.001	−0.67 (0.001)	−0.58 (0.007)
L precentral gyrus	4	−60	−8	27	4.13	0.004	−0.65 (0.002)	−0.54 (0.014)
L middle frontal gyrus	6	−37	4	51	3.33	0.047	−0.81 (0.000)	−0.77 (0.000)
L middle frontal gyrus	8	−24	30	43	3.52	0.028	−0.61 (0.005)	−0.53 (0.016)
9	−46	6	39	3.66	0.018	−0.81 (0.000)	−0.77 (0.000)
L medial frontal gyrus	8	−4	39	40	3.35	0.042	−0.50 (0.026)	−0.45 (0.048)
10	0	61	20	3.55	0.025	−0.60 (0.005)	−0.55 (0.013)
L inferior frontal gyrus	46	−49	40	3	3.42	0.038	−0.61 (0.004)	−0.60 (0.006)
R precentral gyrus	4	45	−12	45	3.70	0.016	−0.79 (0.000)	−0.68 (0.001)
44	52	6	8	3.47	0.032	−0.67 (0.001)	−0.58 (0.007)
R middle frontal gyrus	6	38	−5	56	3.60	0.022	−0.79 (0.000)	−0.68 (0.001)
8	22	31	40	3.53	0.027	−0.61 (0.005)	−0.53 (0.015)
R medial frontal gyrus	11	2	57	−13	3.33	0.042	−0.58 (0.008)	−0.49 (0.028)
R superior frontal gyrus	9	23	40	34	3.39	0.040	−0.62 (0.004)	−0.58 (0.007)
L dorsal anterior cingulate gyrus	32	−1	13	38	3.41	0.038	−0.73 (0.000)	−0.67 (0.001)
R dorsal posterior cingulate gyrus	31	12	−42	43	3.69	0.017	−0.64 (0.002)	−0.65 (0.002)
R inferior parietal gyrus	40	42	−39	47	3.62	0.020	−0.73 (0.000)	−0.67 (0.001)
R superior parietal gyrus	7	26	−63	48	3.46	0.034	−0.73 (0.000)	−0.67 (0.001)
R lingual gyrus	18	1	−79	0	3.45	–	NS	NS

∗

Results of correlation analyses between the voxel values in each significant cluster and the headache duration in years (Pearson correlation test, P < 0.05). The individual P-values are presented in parentheses.

†

Results of correlation analyses between the voxel values in each significant cluster and the estimated lifetime frequency of migraine attacks (Pearson correlation test, P < 0.05).

R, right; L, left; BA, Brodmann's area; SVC, small volume correction; NS, non-significant.

The coordinates refer to the Talairach space and denote the regions of maximal grey matter volume reductions within each cluster (defined as the voxel with the highest Z-value). All regions were significant at P < 0.001, uncorrected for multiple comparisons at a voxel level.

Figure 1

Statistical parametric maps (SPM) demonstrating regional differences in grey matter volume (GMV) between 20 migraine patients and 33 controls. SPM results superimposed on glass brain (a) and standard T1 MRI template images (b–e) show significant GMV reductions in bilateral insula, motor/premotor, prefrontal, cingulate cortex, right posterior parietal cortex, and orbitofrontal cortex (thresholded at P < 0.001, uncorrected for multiple comparisons at a voxel level; corrected P < 0.05 after small volume corrections). The colour bar represents the T-values. The left side of each picture is the left side of the brain.

Discussion

Using VBM, we found that migraine patients had significant GMV reductions in the bilateral insula, motor/premotor, prefrontal, cingulate cortex, right posterior parietal cortex, and orbitofrontal cortex. The regions reported here are currently recognized as being involved in general pain processing or the response to pain (2–7, 16, 18–23). The cingulate cortex and insula have been the most consistently activated regions in PET studies on migraine (3–7) and cluster headache (20, 21), and they are thought to be involved in the emotional/cognitive and autonomic responses to pain, respectively (2, 16, 18). The frontal lobe, especially the premotor and prefrontal cortex, and right posterior parietal cortex are thought to mediate part of the cognitive dimension of pain processing associated with localization and encoding of the attended stimulus (16, 18).

A recent VBM study on migraine patients with T2-visible WM lesions has shown reduced GM density in several brain regions involved in pain processing as well as negative correlations between these GM changes and headache duration (10), which accords well with our findings. The authors suggested that either repeated migraine attacks or retrograde degeneration of axons passing through macroscopic WM lesions might have contributed to the GM changes (10). Our study differed from the previous study in that all our patients had no WM lesions, implying that the observed GM changes seem to be attributed to not only WM lesions but also to repeated migraine attacks per se. In contrast to the previous study (10), however, we could not find any region of GMV increase in the regions specific to migraine (i.e. periaqueductal GM or dorsolateral pons). This discrepancy between the studies cannot be clearly explained, but might be attributed, in part, to the use of different MRI scanners (3 T vs. 1.5 T in ours), because these scanners may have different sensitivities for segmentation of GM/WM and detecting subtle abnormalities. Our findings are in contrast to those of previous VBM studies, which failed to show any GM changes in episodic migraine (8) or medication-overuse headache from chronic migraine (9). One possible explanation for this discrepancy is that migraine is too genetically heterogeneous a disease (24) for investigators to be able to obtain uniform results, despite a phenotypically homogeneous group of patients and the same VBM used in each study. Further VBM studies with a more homogeneous genotype are therefore highly encouraged in order to confirm that migraine patients do in fact have such structural GM changes.

We assume that the GMV changes observed in our patients with episodic migraine may occur as a consequence of repeated brain insults during migraine attacks rather than a specific cause that promotes the chronification of migraine. Considering that the regions found to be activated during migraine attacks (3–7) correspond to the regions of GMV reductions in our results, we speculate that repeated migraine attacks with the continuous activation of the pain-processing regions might result in atrophic changes in these regions. This speculation could be strengthened by our findings of negative correlations between the GMV changes and headache duration or lifetime headache frequency. Moreover, such GM changes have also been demonstrated in other forms of chronic pain disorders, including back pain (22, 25), tension-type headache (9) and phantom limb pain (26). It is of particular interest that patients with chronic tension-type headache, a distinct headache syndrome from migraine, have GM changes similar to those seen in our migraine patients (9). These observations suggest that the observed GM changes are not specific to migraine, but might reflect GM changes as a result of repeated painful attacks in a variety of pain disorders. From this point of view, our findings do not conflict with the recent hypothesis that migraine is primarily a biochemical/biophysical disorder (2, 8).

The possible mechanism underlying the GM changes in migraine is currently unknown. One possibility is that the GM changes might result from repeated ischaemia caused by prolonged reduced perfusion pressure, reduced blood flow, and oligaemia in large and/or small arteries during migraine attacks (27). Another is that local changes during migraine attacks, such as excessive neuronal activation, neurogenic inflammation (28, 29) or excitotoxicity (30), might lead to tissue damage in selected cortical regions. At this point, a question arises whether the GM changes are reversible with proper migraine prophylaxis. It is possible that changes in the extracellular space and microvascular volume may cause tissue shrinkage (GM atrophy) without having a substantial impact on neuronal properties, implying that proper treatment would reverse the atrophic changes of GM (22). The GM atrophy may also be attributed to more irreversible processes, such as neurodegeneration, because most neurodegenerative disorders have shown GM atrophy in VBM studies (31–35). Histopathological studies would correctly answer this question, but it is very unlikely that they would be performed in migraine patients. Further VBM study with a longitudinal design, segregating patients with proper prophylaxis from those without, could be used to shed some light on this issue.

In conclusion, using VBM we have found evidence of structural GM changes in patients with migraine. Our findings of progressive GMV reductions in relation to increasing headache duration and increasing headache frequency suggest that repeated migraine attacks over time result in selective damage to several brain regions involved in central pain processing.

Footnotes

Acknowledgements

The authors thank the participants for taking part in the present study.

References

1 Goadsby

Lipton

Ferrari

. Migraine—current understanding and treatment. N Engl J Med 2002; 346:257–70.

2 May

. A review of diagnostic and functional imaging in headache. J Headache Pain 2006; 7:174–84.

Weiller

May

Limmroth

Juptner

Kaube

Schayck

Brain stem activation in spontaneous human migraine attacks. Nat Med 1995; 1:658–60.

4 Bahra

Matharu

Buchel

Frackowiak

Goadsby

. Brainstem activation specific to migraine headache. Lancet 2001; 357:1016–17.

5 Matharu

Bartsch

Ward

Frackowiak

Weiner

Goadsby

. Central neuromodulation in chronic migraine patients with suboccipital stimulators: a PET study. Brain 2004; 127:220–30.

Afridi

Matharu

Lee

Kaube

Friston

Frackowiak

A PET study exploring the laterality of brainstem activation in migraine using glyceryl trinitrate. Brain 2005; 128:932–9.

7 Afridi

Giffin

Kaube

Friston

Ward

Frackowiak

Goadsby

. A positron emission tomographic study in spontaneous migraine. Arch Neurol 2005; 62:1270–5.

8 Matharu

Good

May

Bahra

Goadsby

. No change in the structure of the brain in migraine: a voxel-based morphometric study. Eur J Neurol 2003; 10:53–7.

Schmidt-Wilcke

Leinisch

Straube

Kampfe

Draganski

Diener

Gray matter decrease in patients with chronic tension type headache. Neurology 2005; 65:1483–6.

10.

Rocca

Ceccarelli

Falini

Colombo

Tortorella

Bernasconi

Brain gray matter changes in migraine patients with T2-visible lesions: a 3-T MRI study. Stroke 2006; 37:1765–70.

11.

11 Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia 1988; 8 (Suppl. 7):1–96.

12.

12 Ashburner

Friston

. Voxel-based morphometry—the methods. Neuroimage 2000; 11:805–21.

13.

13 Kruit

van Buchem

Hofman

Bakkers

Terwindt

Ferrari

Launer

. Migraine as a risk factor for subclinical brain lesions. JAMA 2004; 291:427–34.

14.

14 Headache Classification Subcommittee of the International Headache Society. The International Classification of Headache Disorders, 2nd edn. Cephalalgia 2004; 24 (Suppl. 1):9–160.

15.

15 Good

Johnsrude

Ashburner

Henson

Friston

Frackowiak

. A voxel-based morphometric study of ageing in 465 normal adult human brains. Neuroimage 2001; 14:21–36.

16.

16 Peyron

Laurent

Garcia-Larrea

. Functional imaging of brain responses to pain. A review and meta-analysis. Neurophysiol Clin 2000; 30:263–88.

17.

Lancaster

Woldorff

Parsons

Liotti

Freitas

Rainey

Automated Talairach atlas labels for functional brain mapping. Hum Brain Mapp 2000; 10:120–31.

18.

Peyron

Garcia-Larrea

Gregoire

Costes

Convers

Lavenne

Haemodynamic brain responses to acute pain in humans: sensory and attentional networks. Brain 1999; 122:1765–80.

19.

19 Peyron

Schneider

Faillenot

Convers

Barral

Garcia-Larrea

Laurent

. An fMRI study of cortical representation of mechanical allodynia in patients with neuropathic pain. Neurology 2004; 63:1838–46.

20.

20 May

Bahra

Buchel

Frackowiak

Goadsby

. Hypothalamic activation in cluster headache attacks. Lancet 1998; 352:275–8.

21.

21 May

Bahra

Buchel

Frackowiak

Goadsby

. PET and MRA findings in cluster headache and MRA in experimental pain. Neurology 2000; 55:1328–35.

22.

22 Apkarian

Sosa

Sonty

Levy

Harden

Parrish

Gitelman

. Chronic back pain is associated with decreased prefrontal and thalamic gray matter density. J Neurosci 2004; 24:10410–15.

23.

23 Fumal

Laureys

Di Clemente

Boly

Bohotin

Vandenheede

. Orbitofrontal cortex involvement in chronic analgesic-overuse headache evolving from episodic migraine. Brain 2006; 129:543–50.

24.

24 De Vries

Haan

Frants

Van den Maagdenberg

Ferrari

. Genetic biomarkers for migraine. Headache 2006; 46:1059–68.

25.

25 Schmidt-Wilcke

Leinisch

Ganssbauer

Draganski

Bogdahn

Altmeppen

May

. Affective components and intensity of pain correlate with structural differences in gray matter in chronic back pain patients. Pain 2006; 125:89–97.

26.

26 Draganski

Moser

Lummel

Ganssbauer

Bogdahn

Haas

May

. Decrease of thalamic gray matter following limb amputation. Neuroimage 2006; 31:951–7.

27.

27 Bednarczyk

Remler

Weikart

Nelson

Reed

. Global cerebral blood flow, blood volume, and oxygen metabolism in patients with migraine headache. Neurology 1998; 50:1736–40.

28.

28 Pietrobon

Striessnig

. Neurobiology of migraine. Nat Rev Neurosci 2003; 4:386–98.

29.

29 Waeber

Moskowitz

. Migraine as an inflammatory disorder. Neurology 2005; 64 (Suppl. 2):9–15.

30.

30 Longoni

Ferrarese

. Inflammation and excitotoxicity: role in migraine pathogenesis. Neurol Sci 2006; 27 (Suppl. 2):107–10.

31.

Bozzali

Filippi

Magnani

Cercignani

Franceschi

Schiatti

The contribution of voxel-based morphometry in staging patients with mild cognitive impairment. Neurology 2006; 67:453–60.

32.

32 Burton

McKeith

Burn

Williams

O'Brien

. Cerebral atrophy in Parkinson's disease with and without dementia: a comparison with Alzheimer's disease, dementia with Lewy bodies and controls. Brain 2004; 127:791–800.

33.

33 Price

Paviour

Scahill

Stevens

Rossor

Lees

Fox

. Voxel-based morphometry detects patterns of atrophy that help differentiate progressive supranuclear palsy and Parkinson's disease. Neuroimage 2004; 23:663–9.

34.

Douaud

Gaura

Ribeiro

Lethimonnier

Maroy

Verny

Distribution of grey matter atrophy in Huntington's disease patients: a combined ROI-based and voxel-based morphometric study. Neuroimage 2006; 32:1562–75.

35.

35 Specht

Minnerop

Abele

Reul

Wullner

Klockgether

. In vivo voxel-based morphometry in multiple system atrophy of the cerebellar type. Arch Neurol 2003; 60:1431–5.

Regional Grey Matter Changes in Patients With Migraine: A Voxel-Based Morphometry Study

Abstract

Keywords

Introduction

Methods

Subjects

MRI acquisition and voxel-based morphometry

Statistical analysis

Results

Discussion

Footnotes

Acknowledgements

References