KOA Pathology Screening Using Multi-channel VAG Signal Fusion Method

Knee Osteoarthritis (KOA) is a chronic degenerative lesion of the knee joint, which is difficult to detect clinically. However, the use of knee-joint vibroarthrographic (VAG) signal to assist clinical testing is an effective non-invasive method for KOA pathology diagnosis. In order to improve the correct rate of pathological diagnosis with VAG signals, this paper proposes a KOA pathological screening method using the multi-channel signal fusion method. This method can realize the VAG signal fusion with both high channel consistency and reliability, increase the difference of normal and abnormal VAG signal characteristics realized on the data, improve the KOA pathology screening accuracy by combining time-domain feature extraction and random forest classification methods to obtain features with more discrimination to categories when extracting features. The experimental results showed that this method could obtain high correct rates of KOA pathology screening, with accuracy, precision, and specificity of 0.95, 0.95, and 0.96, respectively, which can be used as an effective method for non-invasive computer-aided KOA diagnosis.