Tecchio F, Vittoria B, Pascarella A, Cottone C, Cancell A, Vitulano D

Introduction: The brain is a connected network, requiring complex-system measures to describe its organization principles [1,2]. Here, we aim at testing whether the normalized compression distance (NCD) [3] is a suitable quantifier of the functional connectivity between cortical regions. This new measure estimates the information shared by two signals comparing the compression length of one signal given the other, without requiring any representation of the single in harmonics or selecting a specific time window where to compare the two signals. We show that this new measure is a good candidate to estimate the inter-nodes connectivity since it displays features 'expected' for brain connectivity, i.e. it is maximal between homologous cortical areas, it is higher for dominant cortical areas, it depends on age. In order to do it we estimated the NCD between functionally homologous primary somatosensory areas (S1) activities, testing the above-mentioned properties. Methods: Twenty-eight healthy, right-handed volunteers participated in the study. We recorded brain magnetic activity in the left and right Rolandic regions by a 28-channel magnetoencephalographic (MEG) system. We recorded rest activity for 3 min in each hemisphere. MEG activity was also collected during the electrical stimulation of the contralateral median nerve at the wrist delivered via surface disks. Elicited electric pulses were 0.2 ms in duration and 631 ms of inter-stimulus interval. Left and right median nerves were separately stimulated, totaling about 200 artifact-free trials for each. We used the Functional Source Separation (FSS) [4,5] algorithm to identify functionally homologous areas in the two hemispheres devoted to the somatosensory hand representation (FS_S1). Therefore, we calculated NCD between the left and right FS_S1s at rest. NCD is a parameter-free, quasi-universal similarity measure, computed from the lengths of compressed data files, singly and in pairwise concatenation. In other terms, NCD defines that two objects are similar if we can significantly "compress" one given the information of the other. We compared the similarity between the left and right homologous areas in single subjects and across the whole group. In particular, we compared the similarity of the activities in the two hemispheres of the same subject, with that in the same or in the opposite hemisphere of different subjects in the group of people. Results: NCD was minimal (maximal functional connectivity) between the neuronal activities of hemispheric functionally homologous areas in the same subject, i.e the NCD between the left and right FS_S1 of the same person was smaller than across different subjects (p<10 -7 consistently). NCD was smaller within the left dominant hemisphere than within the non dominant right one (p=3o10-7), suggesting that more skilled cortical areas express more tuned neuronal activities. Finally, it became more variable in older than younger people (p=.01), indicating that it is sensitive to proprioceptive and sensorimotor skills degradation typical of aging. Conclusions: NCD displayed an excellent ability in quantifying the similarity among neuronal activities, catching the maximal similarity expected for functionally homologous cortical areas of the two hemispheres. It was also sensitive to dominant- and age-dependent properties of somatosensory representation activities. This ability to catch key features of neuronal activity's dynamics indicates NCD as a good candidate for studies of brain functional connectivity, able to overcome the limitations intrinsic to the classical Fourier or autoregressive estimates in assessing the dynamics of two-nodes functional conections.