SEARCH

SEARCH BY CITATION

Keywords:

  • migraine;
  • migraine with aura;
  • visual stimulation;
  • optokinetic stimulation;
  • functional magnetic resonance imaging;
  • functional transcranial Doppler sonography

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References

Objective

This study aimed to assess activation patterns and the hemodynamic response to optokinetic stimulation in migraine with aura patients compared with controls.

Background

It has been proposed that altered visual motion processing in striate and extrastriate visual areas is present in migraine patients and might play a role in the pathophysiology of the disease. Besides activating a large visual network, optokinetic stimulation in particular has been shown to provoke symptoms associated with migraine.

Methods

In this study, we examined the response to visual stimulation in 18 migraine with aura patients compared with 18 healthy controls by using functional magnetic resonance imaging and functional transcranial Doppler, thereby assessing the activation pattern of the visual areas (V1–V5) as well as the vasomotor reactivity of the posterior cerebral artery. For stimulation, we used a vertically rotating optokinetic drum with complex colored figures.

Results

Group analysis of migraineurs with aura vs controls revealed different activation patterns in functional magnetic resonance imaging: attenuation of the physiological right lateralization with a significantly increased activation in the left V5 complex, the left area V3, and the right V5 complex. Analysis of the visually evoked flow response of the cerebral blood flow velocity in the posterior cerebral artery showed a larger side-difference of the offset latency (P < .05) and a reduced steepness of the decreasing slope on the left side (P < .05).

Conclusion

Combining examinations with a good structural (functional magnetic resonance imaging) and temporal (functional transcranial Doppler) resolution is a novel approach to migraine pathophysiology. Our findings of an altered pattern of activation by optokinetic visual stimulation with hyperresponsiveness in visual areas activated by motion perception (V5 and V3) further strengthen the concept of an interictal motion-processing deficit in migraine. This is complemented by the slower restitution of the visually evoked flow response after stimulus offset in migraine with aura patients.


Abbreviations: 
BOLD

blood oxygenation level dependent

CSD

cortical spreading depression

CBFV

cerebral blood flow velocity

FLAIR

fluid attenuated inversion recovery

fMRI

functional magnetic resonance imaging

fTCD

functional transcranial Doppler ultrasound

FEW

family-wise error

GLM

general linear model

LI

laterality index

MA

migraine with aura

MST

middle superior temporal area

MT

middle temporal area

PCA

posterior cerebral artery

SD

standard deviation

TMS

transcranial magnetic stimulation

VEFR

visually evoked flow response

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References

Migraine is a highly prevalent, periodic, and chronic neurological disorder. Substantial research into the pathophysiology has focused on visual processing in migraine patients owing to the fact that typical visual disturbances can occur before and during migraine headache and due to the observation that visual stimuli can trigger an attack of migraine. Experimental data have suggested that the preceding aura symptoms may reflect a cortical spreading depression (CSD) associated with local blood flow changes and transient clinical signs.[1] Assuming that underlying abnormalities are not limited to the attacks, different features of visual processing have been investigated in migraine patients during the interictal phase using transcranial magnetic stimulation (TMS), functional magnetic resonance imaging (fMRI),[2] and functional transcranial Doppler (fTCD).[3] Several of these studies have suggested either a reduced cortical inhibition or an increased neuronal excitability and responsiveness of the primary visual cortical areas, possibly precipitating migraine aura.4-6 In addition, analysis of the dynamic pattern of the cerebrovascular response in migraineurs has led to the assumption that a lack of habituation of the cerebrovascular response might contribute to a disturbance of the metabolic homeostasis of the brain and thereby promote migraine attacks, and may even lead to stroke.[7, 8] While numerous studies have focused on altered visual processing in the striate cortex, recently it has been acknowledged that in migraine patients, changes might also be present at different stages of the visual pathway, including precortical areas as well as higher extrastriate cortical regions.[9] Especially, the latter has been attributed to the deficits in motion processing, such as motion aftereffect prolongation,[10] and a decreased ability to detect coherent motion.[11, 12]

fMRI relies on the measurement of altered metabolic requirements that are influenced by neuronal activity. The highly complex mechanism underlying this process of neurovascular coupling is not completely understood, but is believed to be based on local changes in cerebral blood flow, arterial and venous cerebral blood volume, and cerebral metabolic rate of oxygen utilization, among others. Thereby, fMRI detects neuronal activity indirectly through the blood oxygenation level dependent (BOLD) effect,[13] and offers a detection of brain activation patterns with high spatial resolution. While fMRI has been used to characterize the visual system in healthy humans,[14] similar studies in migraine patients are rare and the results are conflictive. To date, only one previous study has performed MRI with emphasis on interictal motion processing and found that compared with healthy controls, migraine patients showed significantly stronger activation in the left middle-temporal complex.[15]

fTCD assesses vasomotor reactivity within the arterial territory of supply: a task-specific neuronal activation causes an increase in local cerebral blood flow. This method has been applied in migraine patients during visual stimulation showing – among other findings – a greater cerebrovascular response in the middle and posterior cerebral arteries in migraineurs with lesser habituation.[3, 16]

Even though lateralization is a typical clinical feature of migraine, this aspect has been analyzed less frequently in electrophysiological visual processing and cerebral blood flow studies, and the few published studies yielded inconclusive results. While several reports with different techniques did report an interattack asymmetry of findings, a consistent relationship between the side of headache or aura and interictal distribution of functional findings has not been demonstrated.3,17-19

One obvious methodological difference between studies of the visual areas involved in motion perception in migraine patients is the use of a variety of visual stimuli, such as moving dots, rings, gratings, or a black/white checkerboard. As in previous studies,[3, 20] we chose to use a complex, moving, colored visual stimulus resulting in maximum activation of the primary and secondary visual cortices and leading to vertical optokinetic stimulation. Optokinetic stimulation can cause motion sickness due to the mismatch of visual and vestibular perception in otherwise healthy individuals.[21] Migraine sufferers, in turn, have a heightened vulnerability to motion sickness even following milder vestibular stimulation.[22] Because of similarities in symptoms and triggers, it has been postulated that migraine and motion sickness may share a final common pathway; as a recent study showed, optokinetic stimulation can provoke symptoms associated with migraine, especially in patients suffering from this disorder.[23]

Our study was designed to examine the response to visual stimulation with a rotating optokinetic drum in migraine patients compared with healthy controls by using both fMRI with its high spatial resolution and fTCD providing a high temporal resolution, thereby assessing changes in the activation pattern of the visual cortex with attention to extrastriate areas relevant for motion processing as well as in the functional vasomotor reactivity and their lateralization.

Subjects and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References

Population

Eighteen patients (13 women, age range 18–54 years, mean 36.0 ± 11.8) with migraine with aura (MA) according to the diagnostic criteria of the International Headache Society[24] and 18 age- and sex-matched healthy controls with no history of neurological or psychiatric disease (13 women, age range 21–55, mean 36.5 ± 11.4) were recruited for the study. All subjects were right-handed. Patients were studied in a headache-free period, at least 72 hours after the last attack. The local Institutional Review Board (Medical Ethics Committee II, Medical Faculty Mannheim, University of Heidelberg) approved the study. Informed consent was obtained in written form from all subjects. For the subjects’ characteristics, see Table 1.

Table 1. Characteristics of the Study Population
 Migraine With AuraControls
(n = 18)(n = 18)
Female gender, N1313
Mean age, years ± SD36.0 ± 11.836.5 ± 11.4
Age range, years18–5421–55
Mean duration of migraine, years ± SD17 ± 9.25NA
Mean frequency of attacks, per month ± SD3.49 ± 4.68NA
Predominant side of headaches9 left, 4 right, 5 both or no predominant sideNA
Aura characteristics:  
Visual symptoms, N18NA
Speech/language disturbance, N10NA
Sensory or motor symptoms, N10NA
Dizziness and vertigo, N13NA
Patients taking a prophylactic medication, N3NA

Stimulus

The visual stimulus for both fMRI and fTCD examination was a vertically rotating optokinetic drum with complex colored comic figures, as described in previous studies.[3, 20] Subjects were instructed to look at and pursuit the moving images on the drum revolving 10 times per minute.

Data Acquisition

Functional TCD and MRI measurements were conducted consecutively in the same session. As first examination, we performed the trans-temporal TCD recording from the P2 segments of both posterior cerebral arteries (PCA) simultaneously with a four-channel TCD scanner (DWL Multidop X, Compumedics Germany GmbH, Singen, Germany) with 2-MHz pulsed-wave Doppler transducers affixed to a headband during visual stimulation with 10 averaged cycles of a 20-second “on” phase followed by a 20-second “off” phase (eyes closed). During the subsequent fMRI measurement, the stimulus was presented as a movie with the use of the “integrated functional imaging system” (Invivo, Orlando, FL, USA) via a liquid crystal display screen attached to the MRI head coil (see Fig. 1 —). Subjects with visual impairment were provided with MR-compatible corrective lenses. The MR scan was performed on a 1.5-T whole body scanner (Magnetom Sonata, Siemens Medical, Erlangen, Germany). A three-dimensional T1w whole brain data set was acquired (magnetization-prepared rapid gradient echo imaging; repetition time [TR]/echo time [TE]/flip angle/field of view [FOV] = 1.800 milliseconds/3.33 milliseconds/8°/240 mm) for anatomical reference with a voxel size of 1.3 × 1.3 × 1.2 mm3. A fluid-attenuated inversion recovery data set was acquired to rule out pre-existing brain lesions (TR/TE/inversion time [TI]/turbo factor = 9000 milliseconds/108 milliseconds/2400 milliseconds/25; voxel size 0.9 × 0.9 × 5 mm3). For the BOLD fMRI scan, a T2*w echo planar imaging sequence was used (TR/TE/flip angle = 2000 milliseconds/55 milliseconds/90°) with an in-plane resolution of 4 × 4 mm2. Per volume, 20 slices (4 mm thick, 2 mm gap) parallel to the inferior borders of the corpus callosum were scanned in interleaved order. The fMRI run was measured in a blocked design. After 2 ignore measurement volumes that were automatically discarded, 6 baseline blocks of 15 volumes (black screen with fixation cross) altered with 5 task blocks of 10 volumes (rotating optokinetic drum) adding up to a total of 140 volumes (280 seconds).

figure

Figure 1 —. Scheme of the examination procedure: visual stimulation was performed with a vertically rotating optokinetic drum during transtemporal transcranial Doppler recordings from the P2 segments of both posterior cerebral arteries. Exemplary hemodynamic responses are shown. During the subsequent functional magnetic resonance imaging (fMRI) measurement, the rotating drum was presented as a movie via a liquid crystal display screen attached to the MRI head coil, activating all visual areas V1–V5.

Download figure to PowerPoint

Analysis

Data preprocessing, single subject and group analyses were performed using SPM8 (http://www.fil.ion.ucl.ac.uk/spm) implemented in MATLAB (version 7.6.0, The MathWorks Inc., Sherborn, MA, USA). Preprocessing included motion correction, co-registration to the structural images, normalization to the Montreal Neurological Institute 152 brain template and smoothing by an 8 × 8 × 8 mm Gaussian kernel. The first-level single-subject analysis was performed based on the general linear model (GLM) implemented in SPM8. The blocks were convolved with a hemodynamic response function to form task regressors. In addition, the motion parameters were included into the GLM. Second-level mixed-effects analysis was then carried out using the first-level statistic maps. The resulting statistic maps were thresholded at P < .05 using a family-wise error (FWE) correction for multiple comparisons (single-group analyses) or P < .001 (group comparison). Coordinates of activating areas are stated in Talairach space, functional regions were assigned with the SPM anatomy toolbox.[25]

Analysis of the visually evoked flow response (VEFR) of the cerebral blood flow velocity (CBFV) was performed as reported previously, achieving the parameters VEFR relative to the baseline CBFV (VEFR%), onset and offset latency, the off phenomenon, the adaptation, and the steepness of the increasing and decreasing slope.[3] Mean group values and standard deviation (SD) are reported. All parameters were analyzed to identify significant intra-individual side-differences (left side vs right side or vice versa) and between groups of MA patients and controls (side-difference in one group vs side-difference in the other group). A one sample two-tailed t-test was performed concerning a significant side-difference within both groups for all parameters. Side-differences within the groups were tested against each other with independent-samples two-tailed t-test corrected for unequal variances where appropriate.

A laterality index (LI) to characterize hemispheric asymmetry was calculated for fMRI applying the equation LI = (Voxelsright – Voxelsleft)/(Voxelsright + Voxelsleft) and for fTCD using the formula LI = (VEFRright – VEFRleft)/(VEFRright + VEFRleft), a positive LI value representing right hemispheric dominance.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References

None of the patients or the control persons developed headache or motion sickness or reported individual symptoms associated with MA during or immediately after the study.

fMRI

Data analysis of the study population (n = 36) showed a bilateral activation (P ≤ .05; FWE-corrected; total of 24,974 voxels; max T value 17.37) of the striate and extrastriate visual cortex and the lateral geniculate body including complete regions corresponding to V1, V2, V3, V4, and V5 bilaterally.

In the normal control group (n = 18), the main cluster of activation was found in V1, V2, and V3 bilaterally as well as in the right V4 and V5 region (P ≤ .05; FWE-corrected; total of 4544 voxels; max T value 14.48; see Fig. 2 —a). A separate cluster was also identified in the left V4 and V5 regions. Activation pattern showed a significant lateralization (P = .008) to the right hemisphere with a laterality index (SD) of 0.26 (±0.30).

figure

Figure 2 —. The three-dimensional views of group statistical maps of global activation in 18 healthy controls (a) and 18 patients suffering from migraine with aura (b); P ≤ .05, family-wise error-corrected. Group differences in brain activation during visual stimulation between patients and control participants are shown at the bottom panel (c); P < .001, uncorrected. The largest cluster in area V5 is circled in yellow.

Download figure to PowerPoint

In the group consisting of MA patients, the pattern of activation included a cluster corresponding to V1–V5 bilaterally (P ≤ .05; FWE-corrected; total of 11,401 voxels; max T value 22.17; see Fig. 2 —b). Activation in the left hemisphere was more pronounced than in controls; thus, lateralization was significant (P = .02), but less prominent than in the control group with a laterality index (SD) of 0.13 (±0.23). The LI was not significantly different between the groups (P = .168).

Group analysis of MA patients vs controls revealed significantly increased activation in 7 clusters (P ≤ .001 uncorrected; see Table 2). The largest cluster was identified as the left V5 area (118 voxels, coordinates of maximum: –42 –70 10). Other motion sensitive areas activated included the right V5 complex and the left V3 area as well as Brodmann area 7 (precuneus) in the right hemisphere (see Fig. 2 —c). No increased activation was found when comparing controls to MA.

Table 2. Group Difference of Blood Oxygenation Level-Dependent Functional Magnetic Resonance Imaging Effects During Visual Stimulation, P < .001 (Uncorrected)
ClusterRegionPeak Coordinates in Talairach space
SideXYZT value
1 (118 voxels)V5L−42−70104.49
2 (31 voxels)Middle frontal gyrusR4222324.08
3 (29 voxels)Superior occipital gyrusR28−72183.68
4 (22 voxels)V3, area 17, area 18L−16−90−43.74
5 (18 voxels)V5R44−6464.04
6 (14 voxels)Area 7 (7PC), intraparietal sulcus (hlP3, hlP1)R28−48463.58
7 (11 voxels)CuneusL−16−78323.90

fTCD

The characteristics of the responses during stimulation in both groups are summarized in Table 3. Neither the MA nor the control group showed a significant side-difference in the VEFR% or Vmax. Even though not statistically significant, the control group had a higher mean Vmax (54.26 cm/second) than the patient group (49.78 cm/second) and also a higher mean V0 (36.86 cm/second vs 34.73 cm/second). VEFR% in the control group was 47.37% on the left and 49.73% on the right side, while in the MA group, VEFR% was somewhat lower with 42.98% on the left and 45.34% on the right side. Controls as well as patients had higher mean VEFR% values on the right side compared with the left side, however, without statistical significance (laterality index MA 0.04 ± 0.21, controls 0.03 ± 0.12).

Table 3. Parameters of the Visually Evoked Flow Response (VEFR) in Migraine With Aura (MA) and the Control Group
ParameterMean (±SD) MAMean (±SD) ControlsP value
  1. *Significant at P < .05.

  2. L = left; R = right; Δ = difference; units in brackets.

L Vmax (cm/second)49.82 (±12.07)54.27 (±11.74).270
L V0 (cm/second)34.99 (±8.42)37.28 (±9.36).445
R Vmax (cm/second)49.74 (±12.81)54.24 (±7.98).215
R V0 (cm/second)34.47 (±9.78)36.43 (±6.05).474
L VEFR (%)42.98 (±14.70)47.37 (±13.32).355
R VEFR (%)45.34 (±11.07)49.73 (±11.46).250
Δ VEFR (%)11.18 (±10.86)10.35 (±6.66).785
L Onset latency (s)3.48 (±0.59)3.30 (±0.34).265
R Onset latency (s)3.32 (±0.65)3.08 (±0.39).183
Δ Onset latency (s)0.36 (±0.32)0.29 (±0.27).626
L Offset latency (s)5.18 (±1.12)4.74 (±0.50).149
R Offset latency (s)5.41 (±1.27)4.84 (±0.58).097
Δ Offset latency (s)0.51 (±0.37)0.27 (±0.30).042*
L Steepness increasing slope (cm/second2)3.30 (±1.34)3.87 (±1.32).206
R Steepness increasing slope (cm/second2)3.39 (±1.10)3.99 (±1.11).112
Δ Steepness increasing slope (cm/second2)0.78 (±0.72)0.74 (±0.60).851
L Steepness decreasing slope (cm/second2)−3.31 (±1.28)−4.36 (±1.66).041*
R Steepness decreasing slope (cm/second2)−3.77 (±2.32)−4.36 (±1.37).365
Δ Steepness decreasing slope (cm/second2)0.95 (±1.32)1.15 (±1.18).637

The side-difference of the offset latency was significantly larger and the steepness of the decreasing slope on the left side was reduced when comparing the MA vs the control group. This means that the time delay from stimulus offset until a 10% decrease of the PCA flow velocity in MA patients showed a larger difference between the right and left sides than in control subjects (0.51 ± 0.37 second vs 0.27 ± 0.30 second). In controls, the slope of the left PCA flow velocity after stimulus-offset showed a stronger decline compared with the patient group (−4.36 ± 1.66 second vs −3.31 ± 1.28 second).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References

In this study, we used two different techniques – fMRI and fTCD – to assess cerebral hemodynamics in migraine patients during stimulation with a rotating optokinetic drum with complex colored figures and thereby activating striate and extrastriate visual areas involved in motion, pattern, and color perception.

While previous fMRI and TMS studies have suggested an increased cortical reactivity and hyperexcitability in primary visual areas,26-28 more recently the extrastriate visual areas have been identified as a region of differential activation in migraine. Battelli and co-workers were the first to demonstrate a significant difference in the threshold for excitability of bilateral visual areas V5 in persons with migraine using TMS.[4] A robust activation of the area V5 (also known as MT+, hMT+, middle temporal area/middle superior temporal area [MT/MST], or MT/V5+), the human homolog to the medial temporal region in the macaque brain, has been shown by a number of motion stimuli in fMRI studies.14,29-31 Additional studies have highlighted the functional disturbances during visual perception of motion, patterns, and colors in patients suffering migraine,[11, 12, 32] as well as a higher susceptibility to visually induced discomfort or motion sickness.[22, 33, 34] In 2010, Antal et al were the first to describe an increased bilateral activation in the superior-anterior part of the middle-temporal cortex (corresponding to MST) to visual stimulation in migraineurs using a coherent/incoherent moving dot stimulus.[15] Strengthening their findings of extrastriate involvement in migraine, we could not only show significantly increased activation in MA in the motion sensitive cortical areas V5 bilaterally, but also in the left area V3. The area V3 has also been identified as a region related to processing visual motion, possibly as a part of a hypothesized cortical network activity induced by visual motion.[30, 35, 36] However, there is controversy about the exact location and subfields of the V3 complex in humans. Recently, fMRI was used to study the detection of coherent motion vs noise as a measure of global visual motion processing. The authors report greater activation by coherent motion in V5 and putative V3A, but not in V1.[37] Similarly, in another study, the brain activations in areas V2 and V3 to vertical pattern stimuli were significantly higher than to the horizontal pattern stimuli.[38] The greater sensitivity to vertical stimuli has been hypothesized to regulate the preponderance of horizontal visual information.[39] In a different fMRI study design, attention to motion produced a modulation of the activity evoked by moving stimuli in the extrastriate cortical regions V3A and V5.[40]

Studies exploring the interactions between these two motion-sensitive cortical areas and such a cortical network have identified spatial and temporal differences in activation of these regions selective to motion direction and speed.[35, 41] Interestingly, area V3 was suggested to be the source of a CSD-like phenomenon in one subject at the onset of visual aura.[42] These findings suggest an involvement of these functional cortical areas of visual processing in migraine matching structural alterations described in these regions: Granziera et al found increased cortical thickness of motion-processing visual areas V5 and V3 in patients with migraine compared with healthy controls.[43] However, using highly sensitive surface-based morphometry, these findings have recently been challenged.[44]

We also found a larger activation in the right precuneus in migraine patients, an area known to be involved in a number of complex tasks including visuo-spatial imagery and sensitivity to visual motion.[45, 46] However, this finding has to be considered cautiously because of the only small number of voxels detected.

Using fMRI, we found a significant lateralization to the right hemisphere in controls as well as in MA, reflecting a right hemispheric dominance for visuo-spatial and especially optokinetic processing as reported earlier.47-51 Interestingly, the fMRI study of horizontal and vertical optokinetic stimulation in healthy subjects found a right hemispheric predominance in the visual motion-sensitive and ocular motor areas, but not in the primary visual cortex,[47] similar to our findings in the control group. This lateralization was not as distinct in our MA group and in the analysis of the group difference of BOLD fMRI effects during visual stimulation, the largest cluster identified was located in the left hemisphere. These findings suggest a differential processing in extrastriate visual areas in migraine.

As in a previous series of 70 MA patients, we did not find evidence for an increased VEFR% in MA by using fTCD.[3] One possible explanation for this might be the vascular anatomy, as essential parts of the extrastriate visual processing are more likely to be supplied by both the middle cerebral artery and the PCA. While the analyses of our fTCD data also demonstrated higher VEFR% values on the right side compared with the left side – corresponding to the observed right accentuated activation in fMRI – this finding did not reach significant levels. It needs to be stated that in the previous fTCD study, we did find an asymmetry of functional vasomotor reactivity responses from bilateral TCD PCA recordings, with a significant VEFR% side-difference matched with the hemisphere affected by aura symptoms.[3] However, this observation could not be verified in this series, possibly due to the smaller number of patients. Other previous fTCD studies have identified different patterns of neurovascular response to visual stimulation in MA.[16, 52, 53] Our data show a slower restitution of CBFV after stimulus offset in patients compared with controls. This finding might be explained by an excess metabolic demand resulting from the increased activation of the visual areas – complementing the results of our fMRI study.

While fMRI detected differential regional activation patterns with a high spatial resolution, fTCD results depend on a higher oxygen demand in large vascular territories. Metabolically active localized regions at the border of vascular territories – as V5 – might not be adequately depicted. Furthermore, a synchronous recording of bilateral middle and posterior cerebral arteries remains technically challenging. Improvement of the regional resolution of fTCD to assess blood flow changes in distal arterial branches might also help to overcome some of the limitations when comparing the results of these techniques. Concerning fMRI, future investigations of visual motion perception might also include seed-based resting-state fMRI examinations to characterize functionally connected brain networks.

In conclusion, our fMRI findings demonstrate that visual areas activated by motion perception (V5 and V3) are hyperresponsive in MA in the interictal state contributing to the explanation of common interictal motion-processing deficits observed in migraine. Complementary results of fTCD indicate a slower restitution of the hemodynamic response in MA patients.

Statement of Authorship

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References

Category 1

  • (a)
    Conception and Design
    Martin Griebe; Marc E. Wolf; Kristina Szabo; Michael G. Hennerici
  • (b)
    Acquisition of Data
    Martin Griebe; Florian Flux; Marc E. Wolf
  • (c)
    Analysis and Interpretation of Data
    Martin Griebe; Florian Flux; Marc E. Wolf; Kristina Szabo; Michael G. Hennerici

Category 2

  • (a)
    Drafting the Manuscript
    Martin Griebe; Florian Flux; Kristina Szabo
  • (b)
    Revising It for Intellectual Content
    Marc E. Wolf; Michael G. Hennerici

Category 3

  • (a)
    Final Approval of the Completed Manuscript
    Martin Griebe; Florian Flux; Marc E. Wolf; Michael G. Hennerici; Kristina Szabo

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Subjects and Methods
  5. Results
  6. Discussion
  7. Statement of Authorship
  8. References