Abstract
RGB-D cameras provide 3-D body joint data in a low-cost, portable and non-intrusive way, when compared with reference motion capture systems used in laboratory settings. In this contribution, we evaluate the validity of both Microsoft Kinect versions (v1 and v2) for motion analysis against a Qualisys system in a simultaneous protocol. Two different walking directions in relation to the Kinect (towards – WT, and away – WA) were explored. For each gait trial, measures related with all body parts were computed: velocity of all joints, distance between symmetrical joints, and angle at some joints. For each measure, we compared each Kinect version and Qualisys by obtaining the mean true error and mean absolute error, Pearson's correlation coefficient, and optical-to-depth ratio. Although both Kinect v1 and v2 and/or WT and WA data present similar accuracy for some measures, better results were achieved, overall, when using WT data provided by the Kinect v2, especially for velocity measures. Moreover, the velocity and distance presented better results than angle measures. Our results show that both Kinect versions can be an alternative to more expensive systems such as Qualisys, for obtaining distance and velocity measures as well as some angles metrics (namely the knee angles). This conclusion is important towards the off-lab non-intrusive assessment of motor function in different areas, including sports and healthcare. © 2019 Elsevier Ltd

Abstract

Abstract

Abstract
DBS surgery is considered for Parkinson's Disease patients when motor complications and consequent quality of life is no longer acceptable on optimal medical therapy prescribed by neurologists. Within the operating room, the electrode placement with the best clinical outcome for the patient is quantitatively assessed via the wrist rigidity assessment. A subjective scale is used, influenced by the neurologists' perception and experience. Our research group has previously designed a novel, comfortable and wireless system aiming to tackle this subjectivity. This system comprised a gyroscope sensor in a textile band, placed in the patients' hand, which communicated its measurement to a Smartphone via Bluetooth. During the wrist rigidity evaluation exam, a signal descriptor was computed from angular velocity (omega) and a polynomial mathematical model was used to classify the signals using a quantitative scale of rigidity improvement. In this present work, we aim to develop models that consider the 3-gyroscope-axes to acquire the omega and the cogwheel rigidity. Our results showed that y-gyroscope-axis remains the best way to classify the rigidity reduction, showing an accuracy of 78% and a mean error of 3.5%. According to previous results, the performance was similar and the decrease of samples to extract the omega features did not compromise system performance. The cogwheel rigidity did not improve the previous model and other gyroscope-axis beyond the y-axis decreased system performance.

Abstract
Human gait analysis provides valuable information regarding the way of walking of a given subject. Low-cost RGB-D cameras, such as the Microsoft Kinect, are able to estimate the 3-D position of several body joints without requiring the use of markers. This 3-D information can be used to perform objective gait analysis in an affordable. portable, and non-intrusive way. In this contribution, we present a system for fully automatic gait analysis using a single RGB-D camera, namely the second version of the Kinect. Our system does not require any manual intervention (except for starting/stopping the data acquisition), since it firstly recognizes whether the subject is walking or not, and identifies the different gait cycles only when walking is detected. For each gait cycle, it then computes several gait parameters, which can provide useful information in various contexts, such as sports, healthcare, and biometric identification. The activity recognition is performed by a predictive model that distinguishes between three activities (walking, standing and marching), and between two postures of the subject (facing the sensor, and facing away from it). The model was built using a multilayer perceptron algorithm and several measures extracted from 3-D joint data, achieving an overall accuracy and F-1 score of 98%. For gait cycle detection, we implemented an algorithm that estimates the instants corresponding to left and right heel strikes, relying on the distance between ankles, and the velocity of left and right ankles. The algorithm achieved errors for heel strike instant and stride duration estimation of 15 +/- 25 ms and 1 +/- 29 ms (walking towards the sensor), and 12 +/- 23 ms and 2 +/- 24 ms (walking away from the sensor ) Our gait cycle detection solution can be used with any other RGB-D camera that provides the 3-D position of the main body joints.