Adaptive signal processing techniques for analysis of knee joint vibroarthrographic signals

Knee joint articular cartilage diseases such as arthritis affect a significant portion of the elderly. Early noninvasive diagnosis can prevent trauma to the patient, avoid surgery, and reduce health care expenditures. Difficulties exist with imaging and arthroscopy in noninvasively evaluating cartilage surfaces. An interesting possibility of assessing cartilage surface noninvasively is to analyze the vibrations or vibroarthrographic (VAG) signals emitted during normal leg movement.; This thesis explores the diagnostic potential of VAG signals in screening abnormal knees from normal knees via adaptive signal processing techniques. VAG signal characteristics pose severe challenges in terms of analysis and extraction of discriminant features. One of the main characteristics of VAG signals is nonstationarity, which is addressed in the thesis by the use of segmentation-based techniques and time-frequency distributions (TFDs). TFDs overcome the difficulties with segmentation-based methods in relating clinical information with VAG signals.; Novel approaches for denoising VAG signals and construction of adaptive TFDs based on signal decomposition are proposed. The proposed methods are tested with synthetic signals before applying them to real VAG signals. Pattern classification of VAG signal features indicate accuracy up to 86% in screening abnormal knees from normal knees, and has particularly shown high sensitivity in screening patellofemoral articular cartilage disorders such as chondromalacia patella.; An objective method for identifying features in TFDs using an image process' technique in the form of the Hough-Radon transform (HRT) is proposed. The HRT method is tested on TFDs with known time-frequency (TF) dynamics, and is shown to have good potential in automatically identifying TF signatures.; Direct auscultation of knee joints has been a traditional mode of diagnosis. The thesis proposes a computer-aided auscultation technique using auditory display procedures. Subjective evaluation of VAG signals with the proposed sonification technique based on the instantaneous mean frequency has indicated a sensitivity of 83% in screening abnormal knees, although at the expense of a decrease in specificity.; The techniques proposed in the thesis are being incorporated in a diagnostic workstation being prototyped.