A brain-region-based meta-analysis method utilizing the Apriori algorithm. Issue 1 (December 2016)

• Record Type: Journal Article
• Authors: Niu, Zhendong; Nie, Yaoxin; Zhou, Qian; Zhu, Linlin; Wei, Jieyao
• Is Part Of: BMC neuroscience. Volume 17:Issue 1(2016)
• Page Start: 1
• Page End: 7
• Publication Date: 2016-12
• Subjects: Apriori algorithm -- Brain network connectivity -- Co-activation relationship -- fMRI -- Meta-analysis -- Word reading; Neurosciences -- Periodicals; 573.805
• DOI: 10.1186/s12868-016-0257-8
• Languages: English
• ISSNs: 1471-2202

Abstract

Abstract Background Brain network connectivity modeling is a crucial method for studying the brain's cognitive functions. Meta-analyses can unearth reliable results from individual studies. Meta-analytic connectivity modeling is a connectivity analysis method based on regions of interest (ROIs) which showed that meta-analyses could be used to discover brain network connectivity. Results In this paper, we propose a new meta-analysis method that can be used to find network connectivity models based on the Apriori algorithm, which has the potential to derive brain network connectivity models from activation information in the literature, without requiring ROIs. This method first extracts activation information from experimental studies that use cognitive tasks of the same category, and then maps the activation information to corresponding brain areas by using the automatic anatomical label atlas, after which the activation rate of these brain areas is calculated. Finally, using these brain areas, a potential brain network connectivity model is calculated based on the Apriori algorithm. The present study used this method to conduct a mining analysis on the citations in a language review article by Price (Neuroimage 62(2):816–847, 2012). The results showed that the obtained network connectivity model was consistent with that reported by Price. Conclusions The proposed method is helpful to find brain network connectivity by mining the co-activation relationships among brain regions. Furthermore, results of the co-activation relationship analysis can be used as a priori knowledge for the corresponding dynamic causal modeling analysis, possibly achieving a significant dimension-reducing effect, thus increasing the efficiency of the dynamic causal modeling analysis.