Coordinate-based activation likelihood estimation meta-analysis of neuroimaging data: a random-effects approach based on empirical estimates of spatial uncertainty - PubMed (original) (raw)

Coordinate-based activation likelihood estimation meta-analysis of neuroimaging data: a random-effects approach based on empirical estimates of spatial uncertainty

Simon B Eickhoff et al. Hum Brain Mapp. 2009 Sep.

Abstract

A widely used technique for coordinate-based meta-analyses of neuroimaging data is activation likelihood estimation (ALE). ALE assesses the overlap between foci based on modeling them as probability distributions centered at the respective coordinates. In this Human Brain Project/Neuroinformatics research, the authors present a revised ALE algorithm addressing drawbacks associated with former implementations. The first change pertains to the size of the probability distributions, which had to be specified by the used. To provide a more principled solution, the authors analyzed fMRI data of 21 subjects, each normalized into MNI space using nine different approaches. This analysis provided quantitative estimates of between-subject and between-template variability for 16 functionally defined regions, which were then used to explicitly model the spatial uncertainty associated with each reported coordinate. Secondly, instead of testing for an above-chance clustering between foci, the revised algorithm assesses above-chance clustering between experiments. The spatial relationship between foci in a given experiment is now assumed to be fixed and ALE results are assessed against a null-distribution of random spatial association between experiments. Critically, this modification entails a change from fixed- to random-effects inference in ALE analysis allowing generalization of the results to the entire population of studies analyzed. By comparative analysis of real and simulated data, the authors showed that the revised ALE-algorithm overcomes conceptual problems of former meta-analyses and increases the specificity of the ensuing results without loosing the sensitivity of the original approach. It may thus provide a methodologically improved tool for coordinate-based meta-analyses on functional imaging data.

2009 Wiley-Liss, Inc.

PubMed Disclaimer

Figures

Figure 1

Figure 1

Localization of the statistical maxima for the areas listed in Table II identified in the single subject analyses (N = 21). The data shown here is based on normalizing each subjects data into standard stereotaxic space using a linear transformation to the MNI152 template (Lin2MNI152, cf. Table I) and is displayed on a surface view of the MNI single subject template.

Figure 2

Figure 2

Mean Euclidean distance between corresponding maxima (computed pairwise for all possible combination of group‐analyses) shown separately for each region (A: averaged over normalization approaches) and normalization procedure (B: averaged across regions). The average across all areas and normalization s, i.e., the estimate of the between subject variance (EDsub) was 11.6 mm. In A, bars denote the standard deviation across normalization approaches, in B across regions.

Figure 3

Figure 3

Significant results for the contrast “L > 0 ∩ R > 0”, i.e., regions that were considered active independently of the moved hand at a threshold of P < 0.001 (uncorrected). Data is shown for each of the nine group analyses performed in identical fashion on the same subjects but differing in the approaches used to transform each subject's data into MNI space. All results are displayed on a surface view of the MNI single subject template. The juxtaposition of these results illustrates that in spite of identical primary data and statistical processing, the results of fMRI group analyses are indeed influenced by the applied spatial normalization. A more detailed description and in particular a quantitative assessment of these spatial normalization effects on the stereotaxic location of local maxima (equivalent to the between‐template variance in meta‐analyses) are provided in the main text.

Figure 4

Figure 4

Localization of the statistical maxima identified for the areas listed in Table II by the nine group‐analyses using the normalization approaches shown in Table I displayed on a surface view of the MNI single subject template. All group analyses were based on the same data from 21 subjects and performed in the same fashion.

Figure 5

Figure 5

Mean Euclidean distance between corresponding maxima (computed pairwise for all possible combination of group‐analyses) for each of the analyzed brain regions. The average across all areas, i.e., the estimate of the between template variance (EDtemp) was 5.7 mm.

Figure 6

Figure 6

A1: Location of the 24 individual activation foci as reported in an exemplary finger tapping experiment (Gerardin et al.,2000). A2: Summary of the 883 individual activation foci reported in all 73 experiments included in the meta‐analysis (cf. Table III). B1: modeled activation (MA) map resulting from centring Gaussian distributions of 7.02 mm FWHM (based on 8 subjects and estimates of 11.6 mm for EDsub and 5.7 mm for EDtemp) at the location of the foci displayed in A1. B2: Activation likelihood estimates (ALE), reflecting, for each voxel, the union of the MA maps (exemplified in B1) across all experiments. C1: Histogram of the ALE scores obtained in the permutation analysis, i.e., under the assumption of a random spatial association between studies. The values summarized in this histogram hence reflect the null‐distribution against which the experimental ALE scores are compared in order to compute their respective P values. Note, that the empirical null‐distribution is sufficiently smooth even in regions of higher ALE scores to allow a reliable attribution of significance levels to the obtained experimental ALE scores, underlining the benefit of the large numbers of permutations used in their construction. C2: The clusters of significant (P < 0.05, corrected) convergence across studies, i.e., significant voxels from B2 and hence results of the performed meta‐analysis, as assessed by comparison with an empirical null distribution by permutation testing (cf. C1). All data is displayed on a surface view of the MNI single subject template.

Figure 7

Figure 7

Comparison of the results obtained for the coordinate‐based meta‐analysis as performed using the classical ALE algorithm (10 mm FWHM, fixed‐effects inference, no grey‐matter mask) and the new approach outlined in the present paper (FWHM based on empirically derived variance mode, random‐effects inference, grey matter mask). For comparison the results obtained in the present fMRI study for the contrast “L + R > 0” (as also the studies included in the meta‐analysis reported coordinates for the movement of either hand). All data is displayed on a surface view of the MNI single subject template.

Figure 8

Figure 8

Results of the two simulated ALE analyses. The dataset examined in panel (A) consisted of 25 studies (12 subjects each) each featuring one focus in BA 44 and 10 randomly distributed additional foci. Moreover, a single study reported 10 foci in the inferior parietal lobule. Classical ALE analysis indicated significance for both regions (and locations of unintentional convergence between the random foci). In turn, the revised random‐effects approach revealed only the inferior frontal gyrus to have a true convergence between foci from different experiments. The dataset examined in panel (B) consisted of four studies investigating 30 subjects showing well localized foci in BA 44 and 21 studies investigating four subjects featuring more variable foci (each study also contained 10 randomly distributed foci). As hypothesized from the scaling of the FWHM by the sample size, the significant activation was more confined when the revised approach was used, confirming that the revised approach does indeed give higher localizing power to larger and hence more representative studies. As in the first dataset, classical ALE analysis but not the more conservative random‐effects approach also revealed spurious convergence between the random foci.

Similar articles

Cited by

References

    1. Amunts K,Weiss PH,Mohlberg H,Pieperhoff P,Eickhoff S,Gurd J,Marshall JC,Shah NJ,Fink GR,Zilles K ( 2004): Analysis of neural mechanisms underlying verbal fluency in cytoarchitectonically defined stereotaxic space—The roles of Brodmann areas 44 and 45. Neuroimage 22: 42–56. - PubMed
    1. Aoki T,Tsuda H,Takasawa M,Osaki Y,Oku N,Hatazawa J,Kinoshita H ( 2005): The effect of tapping finger and mode differences on cortical and subcortical activities: A PET study. Exp Brain Res 160: 375–383. - PubMed
    1. Aramaki Y,Honda M,Okada T,Sadato N ( 2006): Neural correlates of the spontaneous phase transition during bimanual coordination. Cerebral Cortex 16: 1338–1348. - PubMed
    1. Ardekani BA,Guckemus S,Bachman A,Hoptman MJ,Wojtaszek M,Nierenberg J ( 2005): Quantitative comparison of algorithms for inter‐subject registration of 3D volumetric brain MRI scans. J Neurosci Methods 142: 67–76. - PubMed
    1. Ashburner J,Friston KJ ( 2003) Rigid body registration In: Frackowiak RS,Friston KJ, Frith CD, Dolan RJ, Price CJ, Ashburner J, Penny WD, Zeki S, editors. Human Brain Function, 2 ed. San Diego: Academic Press; pp 635–655.

Publication types

MeSH terms

Grants and funding

LinkOut - more resources