Inertial measurement units (IMUs) are being recognized as a portable and cost-effective alternative to motion analysis systems, and have potential to be introduced into clinical settings for low back pain (LBP) assessment.1 Uncertainties regarding sensor accuracy however is limiting the widespread use and acceptance of IMU-based assessments into routine clinical practice.2 Our lab has developed and is refining a framework for performing IMU-based spine movement quality analyses in clinical settings using a custom mobile application and cloud computing;3 however, prior to implementing the framework on a large scale, it is necessary to assess and validate the performance of the selected sensors for orientation tracking and measurement of spine movement quality.
Ten participants performed 35 cycles of constrained repetitive trunk flexion/extension (FE) at a rate of 15 cycles/minute, while wearing two rigid body reflective marker clusters with IMUs (Mbientlab MetaMotion R sensors; $80USD/unit) attached to the middle of the clusters (Figure 1a,b). Movement data were collected simultaneously from IMUs and Vicon. Three-dimensional angles were calculated in each plane for each cluster, and between clusters to quantify lumbar spine motion. Moreover, local dynamic stability (LDS) was computed on both the FE data and the sum of squares (SS) of the three angles in each plane.3 R-values and ICCs2,1 were used to compare continuous angles and all other outputs (i.e., min, max, range of motion (ROM), and LDS), respectively.
Only select results (i.e., R-values for continuous angles and ICCs for ROM and LDS) from the top cluster are presented (Table 1); however, all followed similar trends. Strong relationships were found in the FE and lateral bend (LB) planes, whereas weaker relationships were seen in the axial twist (AT) plane (Figure 1c).
|Results for top cluster.|
|Direction||ROM (ICC2,1)||Continuous (r)|
Overall, sensor accuracy in measuring absolute orientation is high in both the FE and LB planes, and low in the AT plane. Similarly, the IMU sensors show high correlation with Vicon data in measuring LDS. Higher correlation for LDS in the primary movement plane (i.e., FE) compared to using the SS suggests these inexpensive IMUs are accurate in measuring spine movement quality despite poor off-axis orientation tracking. Future directions include validation in a controlled environment, measurement of additional spine movement quality variables, and further refinement of post-processing methods within the application.