Knee Joint Vibroarthrographic Signal Analysis and Classification Using Wavelet Transform

The knee joint is susceptible to various pathological conditions that alter its biomechanical properties, leading to changes in vibrational characteristics detectable through Vibroarthrographic (VAG) signals. VAG signals present a valuable early diagnostic tool for knee joint disorders, owing to their nonstationary and nonlinear characteristics. Although previous research has developed different time and frequency-based parameters for VAG signal assessment and classification, limited exploration has been devoted to statistical modeling methods in the spectral domain. This study seeks to contribute to the field by classifying VAG signals into normal and abnormal categories, reflecting the health condition of knee joints, through the introduction of a novel set of features derived from wavelet analysis. This study uses a comprehensive approach integrating wavelet decomposition and sub-band feature extraction to analyze VAG signals. It focuses on spectral domain transformations using wavelet transform to extract statistical characteristic features like Energy, Shannon Entropy, Kurtosis, Skewness, and Standard Deviation, which capture fluctuations in VAG signals. The classification task is then performed by using a Support Vector Machine (SVM) and Random Forest algorithms. The proposed system achieves an impressive overall classification accuracy of 93.33% for the Random Forest classifier. This novel approach, leveraging wavelet analysis and statistical features, demonstrates the promising potential for accurate screening, with an ability to effectively classify more than 90% of subjects.