Analysis of knee joint sound signals for non-invasive diagnosis of cartilage pathology.

The need for a safe, objective, noninvasive tool for the early detection, localization, and quantification of both hyaline articular cartilage and meniscal pathology in the knee is discussed, and the possible use of joint sounds for this purpose is examined. A historical survey of joint sound analysis is given, and the authors' own research is described. The analysis of the knee joint sounds, using time-domain signal plots and three-dimensional spectral plots, supported the authors' assumptions regarding the nature of various degrees of chondromalacia and meniscal lesions, and the associated sounds. Quantitative features such as energy, frequency peaks, duration of signal components, and bandwidths can be easily computed from the data. Further subclassification, however, would require more accurate quantification or parametric representation of signal features, which should be possible by modeling techniques such as linear prediction.