Robotically quantifying finger and ankle proprioception: Role of range, speed, anticipatory errors, and learning.

Proprioception plays a key role in motor control and stroke recovery. Robotic devices are increasingly being used to improve proprioceptive assessments, but there is a lack of knowledge about how programmable factors such as testing range, speed, and prior exposure affect tests. From a physiological standpoint, such factors may regulate the sensitivity of limb proprioceptors, thereby influencing assessment results when not controlled for. To determine the relative influence of such factors, we studied the Crisscross proprioceptive assessment, a recently developed robotic assessment that requires participants to indicate when two joints pass by each other as they are moved passively by the robot. We implemented Crisscross with novel robots for the fingers and ankles and tested young unimpaired participants in single sessions (N = 16) and longitudinally (N = 5, across 15-30 sessions over 3-10 weeks). In single-session testing, we found that proprioceptive acuity was better for the fingers than the ankle (p < 0.01). For both limbs, acuity improved near the ends of the range of motion, which may be due to greater involvement of load and joint receptors. Acuity was poorer for slower movements due to greater anticipatory errors. These results show how the range and speed selected for a proprioceptive test affect proprioceptive acuity and highlight the heightened role of anticipatory errors at slow speeds. Improvements in proprioceptive acuity were not detectable in a single session, but acuity improved across multiple testing sessions (p < 0.01). This result shows that multiple prior exposure over at least several days can affect acuity.Clinical Relevance- Proprioceptive assessments should account for range and speed, which could be enabled by leveraging robotics technology. Proprioceptive acuity can be improved through repeated testing, an observation that is relevant to proprioceptive rehabilitation as well.

A Systematic Review of the Learning Dynamics of Proprioception Training: Specificity, Acquisition, Retention, and Transfer.

OBJECTIVE: We aimed to identify key aspects of the learning dynamics of proprioception training including: 1) specificity to the training type, 2) acquisition of proprioceptive skills, 3) retention of learning effects, and 4) transfer to different proprioceptive skills. METHODS: We performed a systematic literature search using the database (MEDLINE, EMBASE, Cochrane Library, and PEDro). The inclusion criteria required adult participants who underwent any training program that could enhance proprioceptive function, and at least 1 quantitative assessment of proprioception before and after the intervention. We analyzed within-group changes to quantify the effectiveness of an intervention. RESULTS: In total, 106 studies with 343 participant-outcome groups were included. Proprioception-specific training resulted in large effect sizes with a mean improvement of 23.4 to 42.6%, nonspecific training resulted in medium effect sizes with 12.3 to 22% improvement, and no training resulted in small effect sizes with 5.0 to 8.9% improvement. Single-session training exhibited significant proprioceptive improvement immediately (10 studies). For training interventions with a midway evaluation (4 studies), trained groups improved by approximately 70% of their final value at the midway point. Proprioceptive improvements were largely maintained at a delayed follow-up of at least 1 week (12 studies). Finally, improvements in 1 assessment were significantly correlated with improvements in another assessment (10 studies). CONCLUSIONS: Proprioceptive learning appears to exhibit several features similar to motor learning, including specificity to the training type, 2 time constant learning curves, good retention, and improvements that are correlated between different assessments, suggesting a possible, common mechanism for the transfer of training.

Concurrent Contribution of Co-contraction to Error Reduction during Dynamic Adaptation of the Wrist.

MRI-compatible robots provide a means of studying brain function involved in complex sensorimotor learning processes, such as adaptation. To properly interpret the neural correlates of behavior measured using MRI-compatible robots, it is critical to validate the measurements of motor performance obtained via such devices. Previously, we characterized adaptation of the wrist in response to a force field applied via an MRI-compatible robot, the MR-SoftWrist. Compared to arm reaching tasks, we observed lower end magnitude of adaptation, and reductions in trajectory errors beyond those explained by adaptation. Thus, we formed two hypotheses: that the observed differences were due to measurement errors of the MR-SoftWrist; or that impedance control plays a significant role in control of wrist movements during dynamic perturbations. To test both hypotheses, we performed a two-session counterbalanced crossover study. In both sessions, participants performed wrist pointing in three force field conditions (zero force, constant, random). Participants used either the MR-SoftWrist or the UDiffWrist, a non-MRI-compatible wrist robot, for task execution in session one, and the other device in session two. To measure anticipatory co-contraction associated with impedance control, we collected surface EMG of four forearm muscles. We found no significant effect of device on behavior, validating the measurements of adaptation obtained with the MR-SoftWrist. EMG measures of co-contraction explained a significant portion of the variance in excess error reduction not attributable to adaptation. These results support the hypothesis that for the wrist, impedance control significantly contributes to reductions in trajectory errors in excess of those explained by adaptation.

Changes in Resting State Functional Connectivity Associated with Dynamic Adaptation of Wrist Movements.

Dynamic adaptation is an error-driven process of adjusting planned motor actions to changes in task dynamics (Shadmehr, 2017). Adapted motor plans are consolidated into memories that contribute to better performance on re-exposure. Consolidation begins within 15 min following training (Criscimagna-Hemminger and Shadmehr, 2008), and can be measured via changes in resting state functional connectivity (rsFC). For dynamic adaptation, rsFC has not been quantified on this timescale, nor has its relationship to adaptative behavior been established. We used a functional magnetic resonance imaging (fMRI)-compatible robot, the MR-SoftWrist (Erwin et al., 2017), to quantify rsFC specific to dynamic adaptation of wrist movements and subsequent memory formation in a mixed-sex cohort of human participants. We acquired fMRI during a motor execution and a dynamic adaptation task to localize brain networks of interest, and quantified rsFC within these networks in three 10-min windows occurring immediately before and after each task. The next day, we assessed behavioral retention. We used a mixed model of rsFC measured in each time window to identify changes in rsFC with task performance, and linear regression to identify the relationship between rsFC and behavior. Following the dynamic adaptation task, rsFC increased within the cortico-cerebellar network and decreased interhemispherically within the cortical sensorimotor network. Increases within the cortico-cerebellar network were specific to dynamic adaptation, as they were associated with behavioral measures of adaptation and retention, indicating that this network has a functional role in consolidation. Instead, decreases in rsFC within the cortical sensorimotor network were associated with motor control processes independent from adaptation and retention.SIGNIFICANCE STATEMENT Motor memory consolidation processes have been studied via functional magnetic resonance imaging (fMRI) by analyzing changes in resting state functional connectivity (rsFC) occurring more than 30 min after adaptation. However, it is unknown whether consolidation processes are detectable immediately (<15 min) following dynamic adaptation. We used an fMRI-compatible wrist robot to localize brain regions involved in dynamic adaptation in the cortico-thalamic-cerebellar (CTC) and cortical sensorimotor networks and quantified changes in rsFC within each network immediately after adaptation. Different patterns of change in rsFC were observed compared with studies conducted at longer latencies. Increases in rsFC in the cortico-cerebellar network were specific to adaptation and retention, while interhemispheric decreases in the cortical sensorimotor network were associated with alternate motor control processes but not with memory formation.

Measurement of stretch-evoked brainstem function using fMRI.

Knowledge on the organization of motor function in the reticulospinal tract (RST) is limited by the lack of methods for measuring RST function in humans. Behavioral studies suggest the involvement of the RST in long latency responses (LLRs). LLRs, elicited by precisely controlled perturbations, can therefore act as a viable paradigm to measure motor-related RST activity using functional Magnetic Resonance Imaging (fMRI). Here we present StretchfMRI, a novel technique developed to study RST function associated with LLRs. StretchfMRI combines robotic perturbations with electromyography and fMRI to simultaneously quantify muscular and neural activity during stretch-evoked LLRs without loss of reliability. Using StretchfMRI, we established the muscle-specific organization of LLR activity in the brainstem. The observed organization is partially consistent with animal models, with activity primarily in the ipsilateral medulla for flexors and in the contralateral pons for extensors, but also includes other areas, such as the midbrain and bilateral pontomedullary contributions.

Training Propulsion via Acceleration of the Trailing Limb.

Walking function, which is critical to performing many activities of daily living, is commonly assessed by walking speed. Walking speed is dependent on propulsion, which is governed by ankle moment and the posture of the trailing limb during push-off. Here, we present a new gait training paradigm that utilizes a dual belt treadmill to train both components of propulsion by accelerating the belt of the trailing limb during push-off. Accelerations require participants to produce greater propulsive force to counteract inertial effects, and increases extension of the trailing limb through increased belt velocity. We hypothesized that one session of training in this paradigm would produce after effects in propulsion mechanics and, consequently, walking speed. We tested the training paradigm on healthy young adults at two acceleration magnitudes-7 m/s2 (HA) and 2 m/s2 (LA)-and compared their results to a third control group (VC) that walked at a higher velocity during training. Results show that the HA group significantly increased walking speed following training (mean ± s.e.m: 0.073 ± 0.013 m/s, p < 0.001). The change in walking speed in the LA and VC groups was not significant (LA: 0.032 ± 0.013 m/s, VC: -0.003 ± 0.013 m/s). Responder analysis showed that changes in push-off posture and in activation of ankle plantar-flexor muscles contributed to the greater increases in gait speed measured in the HA group compared to the LA and VC groups. The duration of after effects post training suggest that the measured changes in neuromotor coordination are consistent with use-dependent learning.

Training propulsion: Locomotor adaptation to accelerations of the trailing limb.

Many stroke survivors suffer from hemiparesis, a condition that results in impaired walking ability. Walking ability is commonly assessed by walking speed, which is dependent on propulsive force generation both in healthy and stroke populations. Propulsive force generation is determined by two factors: ankle moment and the posture of the trailing limb during push-off. Recent work has used robotic assistance strategies to modulate propulsive force with some success. However, robotic strategies are limited by their high cost and the technical difficulty of fitting and operating robotic devices in a clinical setting. Here we present a new paradigm for goal-oriented gait training that utilizes a split belt treadmill to train both components of propulsive force generation, achieved by accelerating the treadmill belt of the trailing limb during push off. Belt accelerations require subjects to produce greater propulsive force to maintain their position on the treadmill and increase trailing limb angle through increased velocity of the accelerated limb. We hypothesized that locomotor adaptation to belt accelerations would result in measurable after effects in the form of increased propulsive force generation. We tested our protocol on healthy subjects at two acceleration magnitudes. Our results show that 79% of subjects significantly increased propulsive force generation following training, and that larger accelerations translated to larger, more persistent behavioral gains.

Identifying the neural representation of fast and slow states in force field adaptation via fMRI.

Although neurorehabilitation is centered on motor learning and control processes, our understanding of how the brain learns to control movement is still limited. Motor adaptation is an error-driven motor learning process that is amenable to study in the laboratory setting. Behavioral studies of motor adaptation have coupled clever task design with computational modeling to study the control processes that underlie motor adaptation. These studies provide evidence of fast and slow learning states in the brain that combine to control neuromotor adaptation.Currently, the neural representation of these states remains unclear, especially for adaptation to changes in task dynamics, commonly studied using force fields imposed by a robotic device. Our group has developed the MR-SoftWrist, a robot capable of executing dynamic adaptation tasks during functional magnetic resonance imaging (fMRI) that can be used to localize these networks in the brain.We simulated an fMRI experiment to determine if signal arising from a switching force field adaptation task can localize the neural representations of fast and slow learning states in the brain. Our results show that our task produces reliable behavioral estimates of fast and slow learning states, and distinctly measurable fMRI activations associated with each state under realistic levels of behavioral and measurement noise. Execution of this protocol with the MR-SoftWrist will extend our knowledge of how the brain learns to control movement.

StretchfMRI: a novel technique to quantify the contribution of the reticular formation to long-latency responses via fMRI.

Increased reticulospinal (RS) function has been observed to cause both positive and negative outcomes in the recovery of motor function after corticospinal lesions such as stroke. Current knowledge of RS function is limited by the lack of accurate, noninvasive methods for measuring RS function. Recent studies suggest that the RS tract may be involved in processing and generating Long Latency Responses (LLRs). As such, LLRs, elicited by applying precisely controlled perturbations, can thus act as a reliable stimulus to measure brainstem function using fMRI with high signal-to-noise ratio.In this paper, we present StretchfMRI, a novel technique that enables simultaneous recording of neural and muscular activity during motor responses conditioned by robotic perturbations, which allows direct investigation of the neural correlates of LLRs.Via preliminary validation experiments, we demonstrate that our technique can reliably elicit and identify LLRs in two wrist muscles-Flexor Carpi Radialis and Extensor Carpi Ulnaris. Moreover, via a single-subject pilot experiment, we show that the occurrence of an LLR in a flexor and extensor muscles modulates neural activity in distinct regions of the brainstem. The observed somatotopic organization is in agreement with the double reciprocal model of RS function observed in animal models, in which the right medullary and left pontine reticular formation are responsible for control of the motor activity in flexors and extensors, respectively.

Quantitative Testing of fMRI-Compatibility of an Electrically Active Mechatronic Device for Robot-Assisted Sensorimotor Protocols.

OBJECTIVE: To develop a quantitative set of methods for testing the functional magnetic resonance imaging (fMRI) compatibility of an electrically-active mechatronic device developed to support sensorimotor protocols during fMRI. METHODS: The set of methods includes phantom and in vivo experiments to measure the effect of a progressively broader set of noise sources potentially introduced by the device. Phantom experiments measure the radio-frequency (RF) noise and temporal noise-to-signal ratio (tNSR) introduced by the device. The in vivo experiment assesses the effect of the device on measured brain activation for a human subject performing a representative sensorimotor task. The proposed protocol was validated via experiments using a 3T MRI scanner operated under nominal conditions and with the inclusion of an electrically-active mechatronic device - the MR-SoftWrist - as the equipment under test (EUT). RESULTS: Quantitative analysis of RF noise data allows detection of active RF noise sources both in controlled RF noise conditions, and in conditions resembling improper filtering of the EUT's electrical signals. In conditions where no RF noise was detectable, the presence and operation of the EUT did not introduce any significant increase in tNSR. A quantitative analysis conducted on in vivo measurements of the number of active voxels in visual and motor areas further showed no significant difference between EUT and baseline conditions. CONCLUSION AND SIGNIFICANCE: The proposed set of quantitative methods supports the development and troubleshooting of electrically-active mechatronic devices for use in sensorimotor protocols with fMRI, and may be used for future testing of such devices.