Pelvis and hip calibration methods for movement analysis with inertial sensors

Abstract

Introduction Inertial measurement units (IMUs) are extensively used today for different clinical applications, gait event detection, fall detection, or physical activity analyse. Compared to the gold standard, the optoelectronic system, IMUs do not require a laboratory environment. Consequently, IMUs seem to constitute the best alternative for movement analysis in natural environment or daily living conditions. IMUs are composed of accelerometers, gyroscopes, and, most often, magnetometers. Data fusion with a Kalman or complementary filter provides the orientation of the IMUs in a global coordinate system. An important prerequisite for human movement analysis is to define the IMU orientation relatively to the human segment on which the IMU is fixed (Picerno 2017), which is achieved by defining the segment axes in the IMU coordinate system. In the literature, different methodologies have been presented. The static calibration method proposes to define the segment axes based on one or several static poses (Vargas-Valencia et al. 2016). The hypothesis is that the segment axes are aligned with the vertical and/or horizontal, depending on the pose taken by the segments. This method is fast to apply but the real body's segment alignment is not ensured. Another method is the anatomical calibration which uses a device to locate in the IMU coordinate system anatomical landmarks on which the axes are based (Picerno, Cereatti, and Cappozzo 2008). This method is repeatable but needs a specific device to locate anatomical landmarks. The last method is the functional method (O'Donovan et al. 2007). The subject performs movements during which the segment rotation axis is defined. This method is also repeatable but is strongly dependant on the subject's ability to perform pure rotation around the functional axis to define. To our knowledge, no functional or anatomical method has been proposed for the pelvis. Now, the pelvis is also an important segment for movement analysis. The pelvis coordinate system definition has indeed an incidence on pelvis kinematics but also on hip kinematics since the hip kinematics are obtained by defining the movement of the thigh coordinate system relatively to that of the pelvis. The aim of the present study is then to compare different calibration methods to obtain the pelvis coordinate system relatively to that of the IMU. We will test a static method, a functional calibration method, and an original method that also makes use of a cali…