Thesis
Evaluating the accuracy of IMU devices in detecting clinically significant changes in knee flexion and extension angles
- Creator
- Rights statement
- Awarding institution
- University of Strathclyde
- Date of award
- 2025
- Thesis identifier
- T17437
- Person Identifier (Local)
- 202163469
- Qualification Level
- Qualification Name
- Department, School or Faculty
- Abstract
- Total knee arthroplasty (TKA) is a widely successful surgical intervention for managing end-stage knee osteoarthritis (KOA), yet patient outcomes are highly dependent on postoperative rehabilitation. Despite this, adherence to rehabilitation programs remains suboptimal, potentially hindering recovery. Wearable inertial measurement units (IMUs) have emerged as promising tools to support rehabilitation and enable early diagnostics of unfavourable recovery through remote monitoring, potentially improving patients’ compliance to rehabilitation protocols and thus improving functional outcomes. However, the clinical utility of these devices depends on their ability to provide accurate measurements of knee joint kinematics, particularly knee flexion angles. This study aimed to evaluate the accuracy of two different wearable IMU devices (a Stryker (USA) commercially available technology, MotionSense™ and a wired IMU research device implementing the Seel Algorithm (Seel, Raisch and Schauer, 2014), in measuring knee flexion angles within clinically significant thresholds. Measurements were evaluated across a diverse healthy adult population of varying ages (20 healthy younger participants, ages ranging between 20 - 36 years old and 14 healthy older participants, ages ranging between 60 - 84 years old) and within a TKA clinical population (10 TKA participants, ages ranging between 53 - 71 years old) both preoperatively and postoperatively (1 week postoperatively and at 6 weeks postoperatively), across a broad range of activities of daily living (ADL’s). The commercially available MotionSense™ technology determines sagittal plane knee angle using a mobile-based app with proprietary software that implements a Madgwick filter (Madgwick, 2010), while the wired research IMU device calculates sagittal plane knee angle using the Seel algorithm (Seel, Raisch and Schauer, 2014). Both technologies’ measurements were compared against the gold standard optoelectronic motion capture system, Vicon, which tracked 16 retro-reflective markers that were attached to the lower body as per the PlugInGaitTM (PIG) model. The methodology used to evaluate the accuracy of each of the IMU devices differed in protocol. Analysis of the MotionSense™ device incorporated a bespoke graphical user interface (GUI) which was used to manually isolated different movement cycles. Following up-sampling to 100Hz using the MATLAB (MathWorks, 2024) interp1 function, cross-correlation was used to time synchronise the movement cycle windows identified from peak flexion to peak flexion using the xcorr MATLAB (MathWorks, 2024) function for each technology. The population mean movement cycle was then analysed for each population group and for each activity, with the pooled mean population range of motion (ROM) assessed. Whereas, following conversion of the raw IMU data into sagittal knee angle measurements using the Seel algorithm (Seel, Raisch and Schauer, 2014), the wired IMU research device data was time synchronised to Vicon data using similar methods, by manually selecting peak knee flexion of each technology. As the sampling frequencies differed between the opto-electronic Vicon motion capture system and the wired IMU research device, Vicon was up-sampled to 200Hz, again by means of interpolation (interp1 function). These measures were then analysed by evaluating each populations mean pooled movement cycle window. For both IMU technologies the zero point for knee flexion depends on marker placement, therefore, the mean knee flexion was subtracted from each data set before calculating a root mean square error (RMSE) between the technologies, determined in each movement cycle window. Results presented RMSE of less than 5° across both devices, across both healthy and clinical populations and across all activities, including those involving larger ROM and higher joint velocities. RMSE values ranged between 0.86° - 4.70° for the MotionSense™ device, while RMSE values ranged between 2.92° - 4.78° for the wired IMU device. No statistically significant differences between the population groups for each technology was evidenced (p > 0.05). Notably, greater discrepancies between the measurement systems were observed during activities involving larger degrees of flexion, for example during the flexion/extension activity performed by the younger healthy population a ROM of 116.5° and RMSE of 3.65° was reported between MotionSense™ and Vicon opto-electronic motion capture system, whereas a RMSE of 1.48° and a ROM of 31.6° was reported for the 1 week postoperative session for the walking activity. Furthermore larger differences were also evidenced during periods associated with faster motion (swing phase displayed larger differences compared to the stance phase for the walking activity). The wearable IMU technologies revealed strong coefficients of correlation and were able to accurately track knee flexion patterns across all population groups. The findings from the TKA cohort underscore the highly patient-specific nature of recovery and postoperative outcomes, further emphasising the need for personalised rehabilitation approaches and the requirement for innovative technologies to deliver this level of personalised care. The use of wearable IMUs within clinical and healthcare settings offers substantial benefits within the recovery period, including remote monitoring capabilities and enhanced compliance with rehabilitation protocols. This study concludes that wearable IMU devices can accurately measure sagittal knee angle supporting their integration into clinical settings. Their ability to provide accurate, objective data validates their use as a practical alternative to traditional in-clinic assessments, particularly in enabling remote and continuous tracking of patient progress. As such, IMUs may represent a valuable asset in modern rehabilitation strategies, facilitating more efficient, patient-centred care.
- Advisor / supervisor
- Riches, Philip
- Resource Type
- DOI
Relazioni
Articoli
| Thumbnail | Titolo | Data caricata | Visibilità | Azioni |
|---|---|---|---|---|
|
|
PDF of thesis T17437 | 2025-10-28 | Pubblico | Scaricare |