A new meta-analytic method for neuroimaging studies that combines reported peak coordinates and statistical parametric maps

Abstract

Meta-analyses are essential to summarize the results of the growing number of neuroimaging studies in psychiatry, neurology and allied disciplines. Image-based meta-analyses use full image information (i.e. the statistical parametric maps) and well-established statistics, but images are rarely available making them highly unfeasible. Peak-probability meta-analyses such as activation likelihood estimation (ALE) or multilevel kernel density analysis (MKDA) are more feasible as they only need reported peak coordinates. Signed-differences methods, such as signed differential mapping (SDM) build upon the positive features of existing peak-probability methods and enable meta-analyses of studies comparing patients with controls. In this paper we present a new version of SDM, named Effect Size SDM (ES-SDM), which enables the combination of statistical parametric maps and peak coordinates and uses well-established statistics. We validated the new method by comparing the results of an ES-SDM meta-analysis of studies on the brain response to fearful faces with the results of a pooled analysis of the original individual data. The results showed that ES-SDM is a valid and reliable coordinate-based method, whose performance might be additionally increased by including statistical parametric maps. We anticipate that ES-SDM will be a helpful tool for researchers in the fields of psychiatry, neurology and allied disciplines.

Mapping the brain correlates of borderline personality disorder: A functional neuroimaging meta-analysis of resting state studies

“…Where only p or z-values were available, we used the online "Two-sample z/p to t converter" available on the SDM website. All the meta-analyses were performed using the anisotropic effect-size-based algorithms (AES-SDM) (Radua et al, 2012;; Radua et al, 2014). The new version of SDM software allows the use of both the positive features of existing peak-probability methods (such as activation likelihood estimation) and of image-based meta-analyses, using standard effect size and variancebased meta-analytic calculations.…”

Coordinate based meta-analysis does not show grey matter atrophy in narcolepsy

“…This reanalyses employed the false discovery rate (FDR) method of controlling the type 1 error [13], which is no longer the recommended option in GingerALE [14] and has been superseded [15]. This has prompted a second re-evaluation of the narcolepsy data by Zhong and colleagues [16] employing the signed differential mapping (SDM) CBMA algorithm [17,18]. The results apparently reaffirm the initial assessment that there is a consistent pattern of grey matter loss in narcolepsy.…”

Age-related parieto-occipital and other gray matter changes in borderline personality disorder: A meta-analysis of cortical and subcortical structures

“…Second, while voxel-based analyses provide excellent control for false positive results, they do not control as thoroughly for false negative results (Radua et al, 2012b). VBM analyses may therefore have insufficient sensitivity to detect between-group differences in small limbic structures (e.g, amygdala, hippocampus) implicated in BPD pathology (Bergouignan et al, 2009).…”

“…Previous simulations showed p ≤ 0.005 (uncorrected) with a cluster-level extent threshold of k ≥ 10 optimally balance false positives and negatives (Radua et al, 2012b).…”

“…In addition to being statistically validated for use in ES-SDM analysis, anisotropic kernels have been successfully applied in other neuroimaging analyses (Van Hecke et al, 2010;Zhang and Davatzikos, 2013;Radua et al, 2014b). This method has been thoroughly reviewed elsewhere (Radua and Mataix-Cols, 2009;Radua et al, 2012b;Radua et al, 2014b) and is only summarized here.…”