Vibration Analysis of Human Knee Joint in Healthy and Osteoarthritic Knees

Abstract

The goal of this thesis is to investigate the possibility of using vibration analysis to detect and assess a very common joint disease known as osteoarthritis (OA). For this purpose, patients with various levels of OA, healthy to severe OA, were recruited and MRI and vibration recordings were made on both knees. MRI images were analyzed by a radiologist and different symptoms related to osteoarthritis in the knee were scored for each observation. Vibration signals of the patients' knees were recorded using 5 accelerometers placed at different locations of the knee. This thesis divides into two major sections; the first section deals with design of an apparatus (a function specific brace and the electronic hardware) for acquiring and recording vibration data from a patient's knee. The second section deals with the analysis of the recorded data using a combination of signal processing techniques (Fourier and wavelet transforms) and multivariate statistical methods (principal component (PCA) and partial least square (PLS)). The brace designed and built for the purpose of this research has several unique properties not found in commercial knee braces. It provides a robust and secure base for attachment of the sensors to the knee and shows very good adaptation to the dynamics of the knee during motion. In the analysis section we show that combining signal processing and multivariate statistical techniques (such as PCA and PLS) provides strong tools for analysis of the data. The result of our analysis shows that there is a strong correlation between vibration analysis and some of the symptoms of osteoarthritis such as cartilage degeneration and formation of osteophytes. We conclude that vibration signals of the knee joint (crepitus) during flexion/extension cycle of the knee, when it is under stress, can be a good indicator of the general severity of OA in patients.