Julien Stamatakis

Finger tapping clinimetric score prediction in Parkinson's disease using low-cost accelerometers

The motor clinical hallmarks of Parkinsons disease (PD) are usually quantified by physicians using validated clinimetric scales such as the Unified Parkinsons Disease Rating Scale (MDS-UPDRS). However, clinical ratings are prone to subjectivity and inter-rater variability. The PD medical community is therefore looking for a simple, inexpensive, and objective rating method. As a first step towards this goal, a triaxial accelerometer-based system was used in a sample of 36 PD patients and 10 age-matched controls as they performed the MDS-UPDRS finger tapping (FT) task. First, raw signals were epoched to isolate the successive single FT movements. Next, eighteen FT task movement features were extracted, depicting MDS-UPDRS features and accelerometer specific features. An ordinal logistic regression model and a greedy backward algorithm were used to identify the most relevant features in the prediction of MDS-UPDRS FT scores, given by 3 specialists in movement disorders (SMDs). The Goodman-Kruskal Gamma index obtained (0.961), depicting the predictive performance of the model, is similar to those obtained between the individual scores given by the SMD (0.870 to 0.970). The automatic prediction of MDS-UPDRS scores using the proposed system may be valuable in clinical trials designed to evaluate and modify motor disability in PD patients.

Development and validation of an accelerometer-based method for quantifying gait events

An original signal processing algorithm is presented to automatically extract, on a stride-by-stride basis, four consecutive fundamental events of walking, heel strike (HS), toe strike (TS), heel-off (HO), and toe-off (TO), from wireless accelerometers applied to the right and left foot. First, the signals recorded from heel and toe three-axis accelerometers are segmented providing heel and toe flat phases. Then, the four gait events are defined from these flat phases. The accelerometer-based event identification was validated in seven healthy volunteers and a total of 247 trials against reference data provided by a force plate, a kinematic 3D analysis system, and video camera. HS, TS, HO, and TO were detected with a temporal accuracy ± precision of 1.3 ms ± 7.2 ms, -4.2 ms ± 10.9 ms, -3.7 ms ± 14.5 ms, and -1.8 ms ± 11.8 ms, respectively, with the associated 95% confidence intervals ranging from -6.3 ms to 2.2 ms. It is concluded that the developed accelerometer-based method can accurately and precisely detect HS, TS, HO, and TO, and could thus be used for the ambulatory monitoring of gait features computed from these events when measured concurrently in both feet.

Brain activation pattern related to gait disturbances in Parkinson's disease†‡§

Gait disturbances represent a therapeutic challenge in Parkinsons disease (PD). To further investigate their underlying pathophysiological mechanisms, we compared brain activation related to mental imagery of gait between 15 PD patients and 15 age‐matched controls using a block‐design functional MRI experiment. On average, patients showed altered locomotion relatively to controls, as assessed with a standardized gait test that evaluated the severity of PD‐related gait disturbances on a 25‐m path. The experiment was conducted in the subjects as they rehearsed themselves walking on the same path with a gait pattern similar as that during locomotor evaluation. Imagined walking times were measured on a trial‐by‐trial basis as a control of behavioral performance. In both groups, mean imagined walking time was not significantly different from that measured during real gait on the path used for evaluation. The between‐group comparison of the mental gait activation pattern with reference to mental imagery of standing showed hypoactivations within parieto‐occipital regions, along with the left hippocampus, midline/lateral cerebellum, and presumed pedunculopontine nucleus/mesencephalic locomotor area, in patients. More specifically, the activation level of the right posterior parietal cortex located within the impaired gait‐related cognitive network decreased proportionally with the severity of gait disturbances scored on the path used for gait evaluation and mental imagery. These novel findings suggest that the right posterior parietal cortex dysfunction is strongly related to the severity of gait disturbances in PD. This region may represent a target for the development of therapeutic interventions for PD‐related gait disturbances.

Gait feature extraction in Parkinson's disease using low-cost accelerometers

The clinical hallmarks of Parkinsons disease (PD) are movement poverty and slowness (i.e. bradykinesia), muscle rigidity, limb tremor or gait disturbances. Parkinsons gait includes slowness, shuffling, short steps, freezing of gait (FoG) and/or asymmetries in gait. There are currently no validated clinical instruments or device that allow a full characterization of gait disturbances in PD. As a step towards this goal, a four accelerometer-based system is proposed to increase the number of parameters that can be extracted to characterize parkinsonian gait disturbances such as FoG or gait asymmetries. After developing the hardware, an algorithm has been developed, that automatically epoched the signals on a stride-by-stride basis and quantified, among others, the gait velocity, the stride time, the stance and swing phases, the single and double support phases or the maximum acceleration at toe-off, as validated by visual inspection of video recordings during the task. The results obtained in a PD patient and a healthy volunteer are presented. The FoG detection will be improved using time-frequency analysis and the system is about to be validated with a state-of-the-art 3D movement analysis system.

Validated extraction of gait events from 3D accelerometer recordings

This work is part of a project that deals with the three-dimensional (3D) analysis of normal and pathological gaits based on a newly developed system for clinical applications, using low-cost wireless accelerometers and a signal processing algorithm. This system automatically extracts relevant gait events such as the heel strikes (HS) and the toe-offs (TO), which characterize the stance and the swing phases of walking. The performances of the low-cost accelerometer hardware and related algorithm have been compared to those obtained by a kinematic 3D analysis system and a force plate, used as gold standard methods. The HS and TO times obtained from the gait data of 7 healthy volunteers (147 trials) have been found to be (mean ± standard deviation) 0.42±7.92 ms and 3.11±10.08 ms later than those determined by the force plate, respectively. The experimental results demonstrate that the new hardware and associated algorithm constitute an effective low-cost gait analysis system, which could thus be used for the assessment of mobility in routine clinical practice.

Study and implementation of a wireless accelerometer network for gait analysis

The purpose of this work is to develop a new wireless system implementing a network of accelerometers, based on the IEEE communication Std. 802.15.4, for gait analysis. 3-axis accelerometers are indeed well adapted for characterizing gait features. The IEEE 802.15.4 standard enables 250 kbps data transfers with very low power consumption. Our network of 10 such RF nodes with 3-axis accelerometers can record accelerations up to ± 10 g and shocks up to 50 Hz for tests on walking and running subjects. Accelerometers have been calibrated according to battery voltage and temperature variations. The average node power consumption is 14 mA, which allows 55 hours of use with two AAA batteries. A base station, connected to a PC, controls the whole 10 nodes system and collects the recorded data. Validation tests have shown that the system handles possible transmission problems. The system is robust and allows real time visualization of accelerations. The system could be easily reconfigured to incorporate other sensors.

Multi-modal movement reconstruction for stroke rehabilitation and performance assessment

Stroke is the leading cause of long-term disabilities in developed countries. 73 to 88% of stroke survivors have altered upper limb function. Currently, these impairments are assessed by clinical tests or via kinematic analysis using expensive commercial visual tracking systems. This paper proposes an hybrid alternative to these costly systems. The tool proposed is based on the one hand on a low-cost camera network i.e. a visual modality, and on the other hand, an accelerometer network i.e. a kinematic modality. The off-line reconstruction quality made with such a system is evaluated by comparison with the Codamotion system reconstruction. Moreover, several kinematic parameters are computed for a set of hemiparetic and healthy subjects executing reach and grasp movements. These features were computed from both systems and compared with mean of correlation factor and random effects models. The low-cost hybrid system demonstrated comparable results with those obtained from the gold standard Codamotion system for all the kinematic features analyzed.

Finger Tapping feature extraction in Parkinson's disease using low-cost accelerometers

The clinical hallmarks of Parkinsons disease (PD) are movement poverty and slowness (i.e. bradykinesia), muscle rigidity and limb tremor. The physicians usually quantify these motor disturbances by assigning a severity score according to validated but time-consuming clinical scales such as the Unified Parkinsons Disease Rating Scale (UPDRS) - part III. These clinical ratings are however prone to subjectivity and inter-rater variability. The PD medical community is therefore looking for a faster and more objective rating method. As a first step towards this goal, a tri-axial accelerometer-based system is proposed as patients are engaged in a repetitive finger tapping task, which is classically used to assess bradykinesia in the UPDRS-III. After developing the hardware, an algorithm has been developed, that automatically epoched the signal on a trial-by-trial basis and quantified, among others, movement speed, amplitude, hesitations or halts as validated by visual inspection of video recordings during the task. The results obtained in a PD patient and an healthy volunteer are presented. Preliminary results show that PD patients and healthy volunteers have different features profiles, so that a classifier could be set up to predict objective UPDRS-III scores.