Reevaluating brain networks activated during mental imagery of finger movements using probabilistic Tensorial Independent Component Analysis (TICA)

The cerebral and cerebellar networks involved in execution and mental imagery of the same sequential finger movements performed with the non-dominant hand were assessed by 3T functional magnetic resonance imaging using multivariate model-free analysis. Eight right-handed healthy volunteers successively performed execution and mental imagery tasks (sequential thumb to fingers opposition). The same data were analyzed by using (1) the linear General Model (p < 0.05 corrected), and (2) probabilistic tensorial independent component analysis (TICA). TICA confirmed that overt movement execution and motor imagery share a common network mainly including: premotor, parietal, insular, temporal, cerebellar cortices and putamen. Motor imagery specifically and bilaterally recruited frontopolar, prefrontal, cingulate, medial insula, neocerebellar cortices and precuneus. Non-dominant hand movements induced bilateral brain and cerebellar activation. In comparison with GLM, TICA identified a more widespread and bilateral network especially during motor imagery. TICA revealed that motor imagery also recruits frontopolar precuneal and occipital cortices, rostral M1/S1 corresponding to the hand somatotopic representation, thalamus and cerebellar lobule VIII. TICA also showed concomitant activation of (1) a cerebello-thalamo-cortical network during motor execution, and (2) a control executive network during imagination. TICA therefore allows precise identification of the brain networks collaborating in the same performance. TICA constitutes a valuable tool to assess and improve detection of brain networks engaged in mental imagery in comparison with GLM.