MULAN: a Multiple Connectivity Analysis Method for Multidimensional Datasets

Huifang Wang 1, @ , Christian Bénar, Pascale Quilichini, Viktor Jirsa, Christophe Bernard, @

1 : Institut de Neuroscience des Systèmes (INS, INSERM U1106)
Inserm
Aix-Marseille Université - Faculté de Médecine, 27, Boulevard Jean Moulin - 13005 Marseille, France - France

Many analysis methods have been introduced to infer the connectivity graph that characterizes the transfer of information between
the recording sites. However, there is no clear consensus on which method can yield the most accurate results under which
conditions. In most cases, a single method is not sufficient to infer the real connectivity graph. In addition, there is no effective
threshold selection method to extract the real connectivity graph from the connection matrices computed by the various analysis
methods. Thus, we propose a MULtiple connectivity ANalysis algorithm, called MULAN, for extracting the most probable
connectivity graph by integrating the results from common used methods. MULAN inference system (MIS) is based on fuzzy logic,
which takes as input the results of the different basic connectivity methods, and produces as output a score for each link in the graph
by applying general rules. We tested MULAN on different types of simulated datasets, taking into consideration graph
configurations, noise levels and connection strengths. And the application of MULAN on real datasets (EEG, FMRI, MEG etc.) is
presented. MULAN is available in an open Matlab toolbox.