Compensatory motor control after stroke: an alternative joint strategy for object-dependent shaping of hand posture

Abstract

Efficient grasping requires planned and accurate coordination of finger movements to approximate the shape of an object before contact. In healthy subjects, hand shaping is known to occur early in reach under predominantly feedforward control. In patients with hemiparesis after stroke, execution of coordinated digit motion during grasping is impaired as a result of damage to the corticospinal tract. The question addressed here is whether patients with hemiparesis are able to compensate for their execution deficit with a qualitatively different grasp strategy that still allows them to differentiate hand posture to object shape. Subjects grasped a rectangular, concave, and convex object while wearing an instrumented glove. Reach-to-grasp was divided into three phases based on wrist kinematics: reach acceleration (reach onset to peak horizontal wrist velocity), reach deceleration (peak horizontal wrist velocity to reach offset), and grasp (reach offset to lift-off). Patients showed reduced finger abduction, proximal interphalangeal joint (PIP) flexion, and metacarpophalangeal joint (MCP) extension at object grasp across all three shapes compared with controls; however, they were able to partially differentiate hand posture for the convex and concave shapes using a compensatory strategy that involved increased MCP flexion rather than the PIP flexion seen in controls. Interestingly, shape-specific hand postures did not unfold initially during reach acceleration as seen in controls, but instead evolved later during reach deceleration, which suggests increased reliance on sensory feedback. These results indicate that kinematic analysis can identify and quantify within-limb compensatory motor control strategies after stroke. From a clinical perspective, quantitative study of compensation is important to better understand the process of recovery from brain injury. From a motor control perspective, compensation can be considered a model for how joint redundancy is exploited to accomplish the task goal through redistribution of work across effectors.

Figures

Fig. 1.

Subjects grasped each object using the whole hand with the thumb on the planar surface and the 4 fingers spread out along and conformed to the shaped surface of the object (A). The extent to which hand posture discriminated the 3 shapes was quantified by the visuomotor efficiency (VME) index. First discriminant analysis was performed to determine whether the hand postures for the 3 shapes were reliably different from each other, by allocating each trial to the shape to which it corresponded most closely in discriminant space, i.e., the space formed by the discriminant functions (B). To detect any overlap in the hand postures for the 3 shapes, the results from the discriminant analyses were used to construct a confusion matrix that summarized the extent to which the hand posture on each trial correctly predicted the shape of the grasped object. Each entry in this matrix represents the number of trials for which the target shape (rows) was predicted by hand posture (columns) (C). If the subject's hand posture matched the target shape at each trial, all entries would be on the diagonal.

Fig. 2.

The average (±SD) abduction, proximal interphalangeal joint (PIP), and metacarpophalangeal joint (MCP) angles of the index, middle, and ring fingers for the rectangular, concave, and convex shapes at the end of grasp in controls (white bars) and patients with hemiparesis (black bars) are shown. Negative numbers represent abduction and extension and positive numbers represent flexion. Note that patients showed reduced abduction and PIP flexion, but increased MCP flexion compared with controls.

Fig. 3.

Representative joint data at grasp from one control and one patient is shown. The data are normalized to the duration of reach-to-grasp. The 2 lines demarcate the 3 phases of the movement based on wrist velocity. Note that the control subject (#4) differentiated the 3 shapes mostly by varying the abduction angles. The PIP excursions of the middle digit also differentiated the concave and convex shapes, but MCP excursions were more similar across all 3 shapes. The patient (#5) was also able to vary the abduction angles across the 3 shapes. The PIP excursions were more similar across the 3 shapes, whereas the MCP excursions appeared to vary instead.

Fig. 4.

The difference in average magnitudes (±SD) of finger abduction, PIP and MCP excursions across the index, middle, and ring fingers for the concave and convex shapes in the 2 groups are shown. While the 2 subject groups modulated abduction angles to object geometry in a similar fashion, controls varied their PIP joints more at grasp, whereas patients varied their MCP joints more.

Fig. 5.

The average VME (±SD) at every 5% interval and the slopes of the VME during reach acceleration, deceleration, and grasp phases of reach-to-grasp are shown. Note that controls showed a steep increase in the slope during reach acceleration, which flattened off during reach deceleration and grasp. In contrast, patients showed a flat VME slope during reach acceleration, which became steeper during reach deceleration and flattened off during grasp.

Fig. 6.

Area plots of the magnitude of the normalized coefficients (all values converted to positive) of the finger abduction, PIP, and MCP angles of the 1st and 2nd discriminant functions for each control subject and patient (represented by the individual colors) at the end of reach deceleration are shown. Note that the total height of the plot is similar for the abduction angles but higher for the PIPs in controls vs. the MCPs in patients. This provides qualitative support for our conclusion that the VME increase during reach deceleration in patients with stroke was driven by an MCP strategy.

Fig. 7.

Correlations between the Fugl-Meyer Scale (FMS) scores and the average index, middle, and ring finger abduction, PIP, and MCP angles for the default rectangular shape at grasp are shown. Patients with lower FMS scores (greater impairment) showed reduced finger abduction and PIP flexion, but increased MCP flexion. These results suggest that individuals with greater levels of impairment compensated by switching over to the MCP strategy.