Motor-imagery ability and function of hemiplegic upper limb in stroke patients

Abstract

Objectives: We quantitatively examined the motor-imagery ability in stroke patients using a bimanual circle-line coordination task (BCT) and clarified the relationship between motor-imagery ability and motor function of hemiplegic upper limbs and the level of use of paralyzed limbs.

Methods: We enrolled 31 stroke patients. Tasks included unimanual-line (U-L)-drawing straight lines on the nonparalyzed side; bimanual circle-line (B-CL)-drawing straight lines with the nonparalyzed limb while drawing circles with the paralyzed limb; and imagery circle-line (I-CL)-drawing straight lines on the nonparalyzed side during imagery drawing on the paralyzed side, using a tablet personal computer. We calculated the ovalization index (OI) and motor-imagery ability (image OI). We used the Fugl-Meyer motor assessment (FMA), amount of use (AOU), and quality of motion (QOM) of the motor activity log (MAL) as the three variables for cluster analysis and performed mediation analysis.

Results: Clusters 1 (FMA <26 points) and 2 (FMA ≥26 points) were formed. In cluster 2, we found significant associations between image OI and FMA, AOU, and QOM. When AOU and QOM were mediated between image OI and FMA, we observed no significant direct association between image OI and FMA, and a significant indirect effect of AOU and QOM.

Interpretation: In stroke patients with moderate-to-mild movement disorder, image OI directly affects AOU of hemiplegic upper limbs and their QOM in daily life and indirectly influences the motor functions via those parameters.

Conflict of interest statement

The authors declare no conflicts of interest associated with this manuscript.

Figures

Figure 1

Bimanual coupling task and an example of a trajectory. (A) Participants performed in three conditions. (1) U‐L condition: participants performed unimanual‐line drawing movements only using the nonparalyzed limb as the control condition. (2) B‐CL condition: participants were asked to draw circles with the paralyzed limb while drawing straight lines with the nonparalyzed limb (bimanual circle‐line condition). (3) I‐CL condition: participants were asked not to move the paralyzed limb but instead to imagine drawing circles with it (imaginary circle‐line condition). For all three conditions, the participants repeatedly drew straight lines on the tablet personal computer using the nonparalyzed limb. (B) On the left side is an example of a trajectory in U‐L condition; on the right side is an example of a trajectory in I‐CL condition. The paralyzed limb's motor image ability was defined as the I‐CL condition OI value minus the U‐L condition OI value (image OI).

Figure 2

Comparison of ovalization index in three conditions. The OI values for the U‐L, B‐CL, and I‐CL conditions were 8.0 ± 3.3%, 11.7 ± 3.3%, and 10.1 ± 4.0% (F = 21.1, P < 0.0001). By postBonferroni correction, the OI value was significantly higher in the B‐CL condition than that in the U‐L condition (P < 0.00003). The OI value was also significantly higher in the I‐CL condition than that in the U‐L condition (P < 0.00003). We did not observe a significant difference between the B‐CL and I‐CL conditions (P = 0.359).

Figure 3

The dendrogram by cluster analysis and the association between FMA and AOU or QOM. (A) The dendrogram is shown on the left side. The orange dotted line divides the patients into two clusters. The cluster centroids for cluster 1 were FMA: 12.50, AOU: 0.10, and QOM: 0.11, and those for cluster 2 were FMA: 46.00, AOU: 1.75, and QOM: 1.64. The cluster sum of squares for cluster 1 was 268.85 and that for cluster 2 was 3,576.66. B: The scatter plot on the right shows the relationship between FMA and AOU (top) or QOM (bottom). The blue dots are cluster 1 and the orange squares are cluster 2. When calculating the association coefficients for FMA with AOU and QOM by cluster, we found significant associations between FMA and both AOU and QOM in cluster 2 (AOU: r = 0.86, S = 214.57, P = 5.56e‐07; QOM: r = 0.90, S = 152.55, P = 2.56e‐08), but not in cluster 1 (AOU: r = 0.216, S = 129.36, P = 0.549; QOM: r = 0.137, S = 142.32, P = 0.705). Cluster 1 showed a floor effect.

Figure 4

The results of mediation analysis. These figures were created using cluster 2 data. The figure above is the result of the AOU model and the figure below is the result of the QOM model. (A) We found a significant positive association between image OI and FMA (standardized β 0.516, SE 0.835, P = 0.017), but no significant association between image OI and FMA (standardized β 0.044, SE 0.680, P = 0.784), although we observed significant positive associations between image OI and AOU (standardized β 0.584, SE 0.104, P = 0.005) and between AOU and FMA (standardized β 0.584, SE 0.104, P = 0.005) with AOU as a mediating variable. (B) We found a significant positive association between image OI and FMA (standardized β 0.516, SE 0.835, P = 0.017), but no significant association between image OI and FMA (standardized β 0.042, SE 0.689, P = 0.796); however, significant positive associations were observed between image OI and QOM (standardized β 0.588, SE 0.097, P = 0.005) and between QOM and FMA (standardized β 0.805, SE 1.312, P < 0.001) with QOM as a mediating variable. The coefficients being displayed are normalization coefficients.