Frequency-domain analysis of IMU data for knee movement disorders following TKA

Enmovi Ltd is affiliated to the established OrthoSensor Inc., a company whose broad aim is to quantify orthopaedics through sensor-assisted technology. Both have recently been taken over by Stryker. Their products aim to enable healthcare providers to deliver evidence-based treatments to improve clinical and economic outcomes for patients and healthcare stakeholders. Enmovi Ltd. have developed a wearable, inertial measurement unit (IMU) technology and associated basic algorithms to monitor knee function of total knee arthroplasty (TKA) patients post-operatively. These data are then presented in a useful form, via an App, to both patient and clinician, informing clinical decision making and motivating the patient to achieve prescribed functional goals. Moreover, ultimately, multi-patient data will be combined to create a big data set in order for machine learning algorithms to identify hitherto unknown connectivities between surgical and post-operative outcomes.

Knee arthroplasty is commonly and successfully employed in the UK for the treatment of knee arthritis. With a growing aged population and increasing prevalence of knee arthritis in these age groups the number of procedures is growing and set to grow further. Revision rates of knee arthroplasty are nationally reported at 3% and pose an additional increasing burden on health services. Optimal post-operative management of TKA patients is key to achieve an overall successful outcome of high patient satisfaction and minimise subsequent referrals and revisions. However, patients do not receive the necessary physiotherapy interventions and struggle to maintain the recommended activities whilst discharged from hospital. A lack of supported rehabilitation often leads to a deficit of function with regards to activities of daily living (ADL) and thus reduced patient satisfaction.

A recent PhD student in the Department of Biomedical Engineering, supervised by the PI, developed a classification algorithm, based on the continuous wavelet transform (CWT) of accelerometer data, which discriminates between those who have self-reported knee instability following TKA, and those who do not. Knee instability following TKA accounts for 15-20% of revisions. Hitherto, no biomechanical description nor quantification of "instability" existed and this work has enabled instability to be more accurately and quantitatively defined than a dichotomous, self-reported, yes/no variable, with significant ramifications for understanding the mechanical aetiology of instability and thus potential interventions. However, this work focussed on one acceleration dimension: the mediolateral direction. An IMU device, which generally includes a gyroscope, enables the effects of gravity to be more easily removed from acceleration data emanating from a lone 3D accelerometer. Thus, IMU data lends itself to a multi-dimensional frequency domain analysis of knee function with enhanced expectations of diagnostic and aetiological performance, not just associated with instability, but potentially other mechanical issues such as aseptic loosening, component malalignment and arthrofibrosis. If demonstrated to be successful, any resulting algorithm can be included in the firmware of the commercial device, providing, for the first time a simple biomechanical, diagnostic test for abnormal knee function following TKA.

Aim
The proposed PhD aims to validate clinical IMU data against gold standard measures; and to develop innovative frequency domain analyses of tri-axial data from the IMU. In doing so, we aim to identify and quantify musculoskeletal pathologies of the TKA knee, and, consequentially, provide mechanistic and aetiological descriptions of these pathologies to inform future clinical interventions.