Quantitative Assessment For Muscle Spasticity Based On SEMG And Motion Sensor

Spasticity is a clinical symptom caused by disordered upper motor neuron,and it is prevalent in patients with stroke,spinal cord injury,cerebral palsy and multiple sclerosis.Quantitative assessment of spasticity was important for clinical management and therapy of these patients.In present,approaches for spasticity assessment can be categorized into three primary ways including clinical scale,biomechanics and electrophysiology.However,these approaches are very likely to have strong subjectivity,device complexity and one-sidedness of evaluation.This study aims to propose a novel method for quantitative assessment of spasticity based on portable sEMG and inertial sensors,to provide objective assessment of spasticity suitable for clinical practice.This study focused on the spasticity of upper arm primarily involving biceps brachii and triceps brachii.The passive elbow stretch under a series of different stretch velocities was performed for subjects,and the stretch reflex activities and elbow passive stretch motion data were recorded by sEMG and inertial sensors.According to the recorded data,the lambda model and kinematic model were constructed.The lambda model was based on the stretch reflex threshold regulated by central neural system,and constructed a linear model between the stretch reflex threshold and the passive elbow stretch velocity,and the intercept of the linear equation was used for spasticity measurement.The kinematic model reconstructed the intended motion curve of the experimenter based on an assumption of constant jerk,and the correlation coefficients between motion curves and the median frequency of angular acceleration curve were extracted for spasticity measurement.These coefficients represented the similarity and stability of motion curve.According to the extracted parameters of the models,two methods were applied for spasticity quantitative measurement based on any single model and model fusion,respectively.The clinical MAS scores from all subjects were regarded as the standard reference,and the mean squared error(MSE)was calculated as the index for evaluating performance of different spasticity assessment methods.The results indicated the method for quantitative assessment of spasticity based on model fusion outperformed any single model.Besides,many methods were applied for quantitative assessment of spasticity based on fusion of model parameters,and the support vector regression with radial basis function kernel achieved the best performance in terms of the lowest MSE after optimization for the hyper-parameters of model.This study proposed a novel approach using fusion of models for objectively evaluating the spasticity for upper limb.The use of portable sEMG and inertial sensors would contribute to wide applications of this approach in clinical,community and home-based rehabilitation practice.The assessment for spasticity could provide accurate feedback about therapy effect of patient,which was valuable for clinician to make a reasonable and effective treatment protocol for patients.