Nadica Miljkovic

Recovery of motor function after stroke: a polymyography-based analysis

We present a method for assessing muscle activation patterns during goal-directed movement. We present a cohort study from a randomized clinical trial that followed the recovery of motor function during and after intensive gait training, assisted by sensor-driven, four-channel electrical stimulation. The instrument that we developed allows for the simultaneous recordings of up to 16 channels that are wirelessly sent to a host computer, which then provides feedback to the subject. The inputs to the portable instrument support electromyography (EMG) amplifiers, inertial sensors and goniometers. We show that this method is sensitive enough to show changes in muscle activation patterns in stroke patients before and after gait training (four weeks, five days a week, 30 min daily). We also show that the recovery decreases the differences between patterns of muscle activities (e.g., levels of muscle activations and median frequencies) assessed in hemiplegic and healthy subjects. This method allows for the analysis of muscle contributions and activation patterns; therefore, it might be possible to better understand the physiology behind the recovery of function. This EMG analysis provides a quantification of recovery that is a valuable addition to other measures, such as the Fugl-Meyer score, the Berg-Balance score, gait speed, and the symmetry index.

Relationship Between Electromyographic Signal Amplitude and Thickness Change of the Trunk Muscles in Patients With and Without Low Back Pain.

Objectives:To compare the relative thickness change of the transversal abdominal (TrA) and lumbar multifidus (LM) muscles during activation in individuals with and without low back pain (LBP), and to establish a relationship between surface electromyography (sEMG) signal amplitude and the relative thickness change of the corresponding muscle during clinically relevant activity, with preferential activation of TrA/LM. Materials and Methods:Thirty-seven pain-free participants and 36 LBP patients were assessed by ultrasound for thickness changes of TrA and LM and by sEMG for changes of electrical activity of the same muscles. sEMG is done with wireless LUMBIA system. The position of the sEMG sensors and activation maneuvers were chosen carefully. Results:Significant group effect was found for relative thickness change of TrA (F1,142=60.69, P<0.0001) and LM (F1,142=36.01, P<0.0001). We found significant correlations between relative thickness change of TrA and sEMG signal amplitude on both sides for LBP (r=0.46 to 0.63, P<0.05) and pain-free patients (r=0.43-0.47, P<0.05). The correlation between LM thickness change and sEMG was significant in pain-free participants for both sides (r=0.36 to 0.38 P<0.05), and right LM in LBP participants (r=0.43, P<0.05), but not for LM in LBP group (r=0.16, P=0.351). Discussion:US and sEMG measurements can be used for objective TrA/LM assessment. Correlation results suggest that the relative change of the muscle thickness could be used as the indicator of the muscle activity. Insight into the activity of TrA/LM in pain-free individuals and LBP patients during and after painful episodes may clarify the role of functional abnormalities of these muscles in LBP.

Independent Component Analysis (ICA) methods for neonatal EEG artifact extraction: Sensitivity to variation of artifact properties

Independent Component Analysis (ICA) is becoming an accepted technique for artifact removal. Nevertheless, there is no consensus about appropriate methods for different applications. This study presents a comparison of common ICA methods: RobustICA, SOBI, JADE, and BSS-CCA, for extraction of ECG artifacts from EEG signal. Algorithms were applied to the data created by superimposing artifact free real-life neonatal EEG and synthetic ECG. Their sensitivity to variation of noise property was compared: we examined variability of Spearman correlation coefficients (SCC) for various Heart Rates (HR) in each of ICA methods. Results show that SOBI and BSS-CCA methods were less sensitive than RobustICA and JADE to artifact alterations (mean SCCs were 0.85 and 0.85 compared to 0.80 and 0.73, respectively) being quite successful in source signal extraction.

Pulse rate assessment: Eulerian Video Magnification vs. electrocardiography recordings

Non contact pulse rate assessment is of interest for patients with cardio-vascular diseases. We tested the feasibility of the use of regular web camera for pulse rate assessment. The Eulerian Video Magnification was used for processing of acquired video sequences. The results were compared with the pulse rate determined from the electrocardiography signals, recorded from the two electrodes positioned at left side of the chest. We present pulse rate results obtained from both methods recorded from two healthy volunteers. The comparison shows that the Eulerian Video Magnification differs from the electrocardiography for less than 5% at normal pulse rate. The results show that for heart rates that correspond to cardio-vascular diseases (e.g. tachicardia, bradicardia), the algorithm needs further modifications (higher quality camera is needed).

H-reflex recorded by multi-pad EMG electrodes

We recorded H-reflex using 4 × 4 multi-pad EMG electrodes on muscle Soleus from 4 healthy subjects. The results indicate that the new proposed method of recording H-reflex provides more information than standard bipolar or unipolar EMG recording, since it enables spatial and temporal parameters assessment. We have developed an algorithm which can automatically select the H-reflex with maximum peak to peak amplitude. We also examined the H-reflex propagation by the use of EMG topography map based on the latency.

sEMG-based detection of poor posture: A feasibility study

The cost of the medical treatment of low back pain (LBP) was estimated to be

Clustering Technique for Quantitative Assessment of Motor Function in Stroke Patients

24 billion in the early 90s. Also, 80% of the LBP is estimated to be due to poor or inappropriate posture. The ultimate goal of the project is to develop a surface electromyography (sEMG)-based device that could be used to prevent and treat LBP by postural re-education or simply for on-the-spot sEMG feedback. In this paper we present the results and conclusions of a feasibility study for sEMG-based poor posture classifier. The results show that a s-EMG based poor posture classifier could be possible. The sensitivity for the best linear classifier model was 72% and the specificity was 78%. The same signal feature returned very different results from one participant to another. This inter-subject variability could be due to different muscular activation patterns during posture correction.

Posture in dentists: Sitting vs. standing positions during dentistry work - an EMG study

We designed a quantification method for assessment of motor function in stroke patients. The method is based on clustering of data from polymyography (pEMG) recordings of the lower leg. We recorded pEMG in healthy subjects and hemiplegic patients during well-controlled single joint voluntary movement – ankle dorsiflexion. Agglomerative Hierarchical Clustering (AHC) method is applied on relative contribution parameters of muscles during dorsiflexion. AHC method differentiated three clusters: cluster of stroke patients at the therapy onset, cluster of stroke patients after Functional Electrical Therapy (FET) therapy, and cluster of healthy subjects, when applied on the parameters obtained from tibialis anterior muscle and rectus femoris muscle. The presented results indicate that the applied clustering technique might be used as a means for studying muscle activation patterns.

Recording and assessment of evoked potentials with electrode arrays

INTRODUCTION Adequate working posture is important for overall health. Inappropriate posture may increase fatigue, decrease efficiency, and eventually lead to injuries. OBJECTIVE The purpose was to examine posture positions used during dentistry work. METHODS In order to quantify different posture positions, we recorded muscle activity and positions of body segments. The position (inclination) data of the back was used to assess two postures: sitting and standing during standard dental interventions. RESULTS During standard interventions, whether sitting or standing, a tilt of less than 20 degrees was most prevalent in the forward and lateral flexion directions. Amplitude of electromyography signals corresponding to the level of muscle activity were higher in sitting compared with the electromyography in standing position for all muscle groups on the left and right side of the body. Significant difference between muscle activity in two working postures was evident in splenius capitis muscle on the left (p = 0.032), on the right side of the body (p = 0.049) and in muscle activity of mastoid muscle on the left side (p = 0.029). CONCLUSION These findings show that risk for increased fatigue and possible injures can be reduced by combining the sitting and standing occupational postures.

Assisting persons after stroke to restore gait: hybrid system

Abstract In order to optimize procedure for the assessment of evoked potentials and to provide visualization of the flow of action potentials along the motor systems, we introduced array electrodes for stimulation and recording and developed software for the analysis of the recordings. The system uses a stimulator connected to an electrode array for the generation of evoked potentials, an electrode array connected to the amplifier, A/D converter and computer for the recording of evoked potentials, and a dedicated software application. The method has been tested for the assessment of the H-reflex on the triceps surae muscle in six healthy humans. The electrode array with 16 pads was positioned over the posterior aspect of the thigh, while the recording electrode array with 16 pads was positioned over the triceps surae muscle. The stimulator activated all the pads of the stimulation electrode array asynchronously, while the signals were recorded continuously at all the recording sites. The results are topography maps (spatial distribution of evoked potentials) and matrices (spatial visualization of nerve excitability). The software allows the automatic selection of the lowest stimulation intensity to achieve maximal H-reflex amplitude and selection of the recording/stimulation pads according to predefined criteria. The analysis of results shows that the method provides rich information compared with the conventional recording of the H-reflex with regard the spatial distribution.