Viscous field training induces after effects but hinders recovery of overground locomotion following spinal cord injury in rats

Abstract

Robotic-assisted gait training was able to improve the unassisted overground locomotion of rats following a cervical spinal cord injury. Specifically, four weeks of daily step training in the Robomedica Rodent Robotic Motor Performance System, where the device actively guided the hindlimbs through a pre-injury stepping pattern while the rats walked over a moving treadmill belt in a quadrupedal posture, was able to improve unassisted overground locomotion as measured by the CatWalk gait analysis device. Unfortunately the improvements were minimal. In fact, control animals that received only body weight supported treadmill training and no active robotic forces showed an even greater restoration of unassisted overground locomotion. This led us to further investigate the effects of the specific forces used in rehabilitative training. The robotic training device was modified to apply assistive (negative viscosity) or resistive (viscous) fields in lieu of the standard active guidance. Within the device, daily training with a viscous field resulted in small, constrained steps that were similar to pre-injury steps. However, when the robot was off for weekly assessments, the steps opened up and deviated away from pre-injury levels. Training in a negative viscosity field produced the opposite effect; large open steps that were unlike pre-injury during daily training, and constrained steps that were more like pre-injury during weekly assessment. These training induced after-effects washed out 2 weeks after the cessation of training. Additionally, these distinct after effects seen in the training device did not translate to distinct differences in the recovery of unassisted overground locomotion, with the body weight supported treadmill training controls showing the greatest recovery of overground locomotion. Still, the fact that different applied forces can induce different after effects has interesting implications for rehabilitative training - is it better to have healthy looking steps during training only to induce abnormal after effects, or have abnormal performance during training but with desirable after effects? The data presented here is the first step in addressing this question.

Keywords: Negative viscosity; Robotic gait training; Spinal cord injury.

Fig 1.. Robot assisted gait training

Robotic actuators attached to hindlimbs (above the ankle) record position and/or apply training forces. The cantilever beam provides variable body weight support while rats walk on the treadmill.

Fig 2.. Mean forces applied to left hindlimb during training.

Multidimensional analysis shows the mean forces applied (red arrows, every 2% of gait cycle) as the steps progress in x,y,t space. +Viscous (violet, upper left) and −Viscous (neon violet, upper right) trained groups produce generally predictable opposing force profiles – large forces during early swing, smaller forces during stance. Active guidance (aqua, bottom) to match a pre-injury steeping pattern (purple line) produces a less predictable force profile. Forces (in Newtons) are scaled to fit within graphic, with the forces in the Active group much higher than the +/−Viscous groups. For all graphs initial contact is at the origin with swing phase negative time, and stance phase positive time (swing phase is also a darker shade). All data taken from the first 2 min of the training session on the 3rd day of the 4th week (week 4.5).

Fig 3.. Robotically assisted gait training induces after effects.

Colored meshes are from the right hindlimb week 4.5 training session and gray meshes are from the week 5 assessments (no BWS or robotic forces, +24hrs after last training session). The Null (navy, upper left) group shows the influence of just BWS. The Active (aqua, upper right) group shows that animals can retain what they learned in training, but are less consistent (similar meshes, but different volumes). The +/− viscous fields show the ability to induce aftereffects. Training in a viscous field (violet, lower left) results in a constrained step pattern that opens up after training, whereas training in a negative viscosity field (neon violet, lower right) results in an open step pattern that constrains after training. For all graphs initial contact is at the origin with swing phase negative time, and stance phase positive time (swing phase is also a darker shade).

Figure 4.. RAGT does not improve stepping within training device.

A) In the Untrained animals (umber) neither limb recovers, and the less impaired left hindlimb spontaneously deviates further away as the weeks progress. This is interpreted as the development of a compensatory strategy. When animals receive daily BWSTT (Null –navy) the less impaired left limb deviates further away from pre-injury where the right limb improves stepping during the active training sessions (gray dots between weeks) but induces a deleterious after-effect when the BWS is removed for weekly assessments (open markers). Actively guiding the hindlimbs through a stepping pattern (Active – aqua) results in right hindlimb stepping patterns that are even less like healthy steps, but reduces the compensatory strategy observed in the Null group. B) While training in a viscous field (+Viscous-violet) both hindlimbs exhibited stepping patterns that were very different from healthy steps. However, during assessment an aftereffect was observed where stepping patterns were more like healthy steps. The opposite is true of training in a negative viscosity field (−Viscous -neon violet). C) All training techniques employed performed similarly to the untrained animals with neither limb returning to pre-injury levels at the end of training (5 weeks) or 2 weeks after the cessation of training (7 weeks). When the two limbs are summed together it can be seen that at the end of training the +Viscous group performed the best followed closely by the Untrained animals. Two weeks after the cessation of training the performance of the animals is slightly different with the Untrained animals the best, and −Viscous a close second. Relative error greater than 2.18 is considered meaningfully different and denoted with a solid gray line.

Figure 5.. BWSTT improves overground locomotion more than RAGT.

A) 1 week following injury, before any training, the unassisted overground locomotion is meaningfully different from pre-injury. Over the next 7 weeks the Untrained (umber) animals spontaneously deviate further away from pre-injury locomotion. Weekly BWSTT (Null-navy) slightly reduces this difference, and this improvement is maintained for two weeks after the cessation of training (weeks 5-7). Weekly RAGT where the hindlimbs are guided through a pre-injury stepping pattern (Active-aqua), is better than untrained animals, but not as good as Null. B) Weekly training in a viscous (+Viscous-violet) or negative viscosity field (−Viscous- neon violet) does not improve recovery of unassisted overground locomotion. Additionally, there is no consistent pattern to indicate a wash-out effect after the cessation of training at week 5. C) After 4 weeks of training (week 5), and 2 weeks after the cessation of training (week 7), BWSTT without any forces applied to the hindlimbs (Null-navy) is the least different from pre-injury. Relative error greater than 2.18 is considered meaningfully different and denoted with a solid gray line.