High-frequency ultrasound backscatter analysis of hyaline cartilage

Abstract

High-frequency ultrasound is a promising method to characterize osteoarthritic changes of cartilage tissue in vivo. Numerous studies have investigated the cartilage surface reflection and successfully related derived ultrasound surface parameters to the degeneration stage of cartilage. Only a few studies evaluated acoustic backscatter originating from the cartilage matrix and so far only integrated spectral amplitudes of the received signals have been considered, particularly the apparent integrated backscatter (AIB). However, information contained in the ultrasound frequency domain could potentially exhibit higher sensitivities towards degenerative changes at a cellular level and therefore provide additional information that may enable a better discrimination between degeneration stages. The aims of this work were to introduce parameters that relate to the frequency dependence of ultrasound backscatter and envelope statistics, to study their diagnostic value in comparison with the well-established AIB, to attempt to classify different degeneration stages using combinations of ultrasound parameters and finally to study the contributions of collagen, proteoglycan and chondrocytes to ultrasound backscatter amplitude. Results showed that the introduced apparent frequency dependence of backscatter (AFB) was sensitive to degenerative changes, particularly the earliest signs of degeneration, whereas the conventionally used AIB exhibited relatively low sensitivities. Furthermore, it was demonstrated that in a classification-based approach, combinations of ultrasound parameters can be used to discriminate between different, particularly very early, degeneration stages. Here, backscatter frequency dependence and backscatter amplitude appeared as good predictors for the discrimination of early and advanced degeneration stages, respectively. Finally, it was shown that more than 50% of the backscatter amplitude can be attributed to natural variations of cell number density and collagen concentration. This work demonstrated for the first time that envelope statistics and depth-dependent spectral slope parameters are highly sensitive to the early stages of extracellular cartilage matrix degeneration and outperform conventionally used amplitude-based parameters. Moreover, these parameters provide beneficial additional information in a classification approach to discriminate between different degeneration stages. The identification of collagen and chondrocytes as major scattering sources at higher ultrasound frequencies suggests that the observed variations relate to degenerative changes of chondrocytes and the collagen network. An application of these parameters and particularly the use of parameter combinations to intra-articular ultrasound arthroscopies appears feasible at present and likely improves the diagnostic potential of these examinations.