An fMRI study on the generalization of motor learning after brain actuated supernumerary robot training

Abstract

Generalization is central to motor learning. However, few studies are on the learning generalization of BCI-actuated supernumerary robotic finger (BCI-SRF) for human-machine interaction training, and no studies have explored its longitudinal neuroplasticity mechanisms. Here, 20 healthy right-handed participants were recruited and randomly assigned to BCI-SRF group or inborn finger group (Finger) for 4-week training and measured by novel SRF-finger opposition sequences and multimodal MRI. After training, the BCI-SRF group showed 350% times compared to the Finger group in the improvement of sequence opposition accuracy before and after training, and accompanied by significant functional connectivity increases in the sensorimotor region and prefrontal cortex, as well as in the intra- and inter-hemisphere of the sensorimotor network. Moreover, Granger Causality Analysis identified causal effect main transfer within the sensorimotor cortex-cerebellar-thalamus loop and frontal-parietal loop. The findings suggest that BCI-SRF training enhances motor sequence learning ability by influencing the functional reorganization of sensorimotor network.

Fig. 1. The behavioral performance of two groups.

The accuracy rates are reported at Pre, Post and FU points. a The changing trend of control sequence accuracy rate (CS). b The changing trend of training sequence accuracy rate (TS).

Fig. 2. The composition of the sensorimotor network.

The visual depiction of the regions of interest (ROIs) of sensorimotor network after task-based fMRI data analysis. BG, basal ganglia; THA, thalamus; Cere, cerebellum, S1M1, the primary sensorimotor cortex; SMA, the supplementary motor area; L corresponds to the left hemisphere, and the R corresponds to the right hemisphere.

Fig. 3. Compared to the finger group, the BCI group showed a significant increase in functional connectivity between the primary sensorimotor cortex and the middle frontal gyrus.

Significant increase in FC between S1M1_L and Frontal_Mid_L. a Seed-based whole brain FC found the Frontal_Mid_L was the significant change area when seed was S1M1_L. The color-coded are illustrates the significant clusters of Frontal_Mid_L. b A two-way repeated measure ANOVA revealed a significant increase of FC between S1M1_L and the Frontal_Mid_L for the BCI group (red) as compared to the finger group (light blue). FC: functional connectivity. S1M1: the primary sensorimotor cortex; L corresponds to the left hemisphere, and the R corresponds to the right hemisphere.

Fig. 4. Changes in inter- and intrahemispheric FC of the sensorimotor network and their correlation with behavioral performance.

BCI-SRF group exhibited a significant increase in inter- and intra-hemispheric FC of the sensorimotor network, and these increases were significantly correlated with the changes of CS accuracy rate. a Comparison of intrahemispheric FC in the left hemisphere between the BCI-SRF group and the Finger group at pre- and postintervention. b Comparison of intrahemispheric FC in the right hemisphere between the BCI-SRF group and the Finger group at pre- and postintervention. A two-way repeated measure ANOVA revealed a significant change between two groups. c Comparison of interhemispheric FC between the BCI-SRF group and the Finger group at pre- and postintervention. A two-way repeated measure ANOVA revealed a significant change between two groups. d Significant correlation was found between the change of intrahemispheric FC in the right hemisphere and control sequence accuracy rate (CS) change. e Significant correlation was found between the change of interhemispheric FC and control sequence accuracy rate (CS) change. FC: functional connectivity; CA: control sequence accuracy rate; Error bars indicate SD.*p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 5. The FC changes and causal effect between the seed points of the sensorimotor network.

The BCI-SRF group showed enhanced FC in the cortical and subcortical brain regions and increased causal effects in the frontal-parietal circuit and the cortical-subcortical circuit. a Comparison of each pair FC between the BCI-SRF group and the Finger group at pre- and postintervention. A significant TIME × GROUP interaction effect was found between these 4 connections. (*p < 0.05, **p < 0.01, ***p < 0.001 see text for results). b The effective connectivity was visualized with the BrainNet Viewer. BG, basal ganglia; Cere, cerebellum; THA, thalamus; S1M1, the primary sensorimotor cortex; SMA, the supplementary motor area. L corresponds to the left hemisphere, and the R corresponds to the right hemisphere. c A model showing the distinct causal connectivity of frontal-parietal (SM_cortex) circuit and the SM_cortex-cerebellar-thalamus cortical-subcortical circuit between the two groups. “+/-” represents BCI-SRF group was significant higher/lower than finger group in causal connectivity, and the arrow indicates the direction of connection. SM_cortex: the integration of S1M1_R and SMA_L; THA: thalamus; Cere: cerebellum; S1M1: the primary sensorimotor cortex; SMA: the supplementary motor area; L corresponds to the left hemisphere, and the R corresponds to the right hemisphere.

Fig. 6. The introduction to training system and training paradigm.

The working principle and display of BCI-SRF after wearing, as well as the motor imaginary guidance interface and training and control sequence during the testing phase. a The composition of the wearable hardware system of the BCI-actuated supernumerary robotic finger (SRF). The brain-actuated SRF motion control efferent pathway and the SRF tactile sensory afferent pathway form an extra SRF control circuit. Furthermore, innate limb motor circuit for the finger movement was also displayed. Both circuits are schematic and do not represent actual neurotransmission pathways. (The person gave permission for the use of his image). b The sequence of motor imagery paradigm. c The display of the behavioral measure paradigm.