The overall goal of this proposed research is to develop and test new ultrasound backscattering models for future quantitative-ultrasound (QUS) -based in-vivo screening and monitoring of early osteoarthritis (EOA). Osteoarthritis (OA) is a joint disease that degenerates articular cartilage (AC) and is the most-prevalent joint disease in the United States causing more than $60 billion in health-care costs each year. Currently, none of the available non-invasive modalities is capable of assessing the early, symptomless stages of OA. By the time symptoms become apparent, OA is usually advanced and cannot be reversed or halted, which limits effective treatment. Therefore, a non-invasive tool that is capable of detecting early signs of cartilage degradation, such as chondrocyte apoptosis, before the patient recognizes symptoms would lead to a paradigm shift in managing osteoarthritis. Studies performed during the past decade indicate that QUS has great potential as such a tool. Several spectral parameters from backscattered ultrasound have been shown to be sensitive to morphological properties of articular cartilage that are related to early developmental stages of OA including cartilage-matrix and cell-morphology parameters. In particular, our most-recent studies indicate that the structural organization of hyaline cartilage in the knee joint is causing coherent scattering that may have a significant impact on QUS- parameter estimation. We will develop a novel and accurate model of the ultrasound backscatter coefficient (BSC) in human articular cartilage. The model will be used to develop a QUS-based, multi-feature approach for classifying articular cartilage and detecting EOA stages and can be implemented in current clinical scanners. We will use a combination of numerical ultrasound simulations and ex vivo measurements to develop the new BSC models and to define the optimal set of QUS parameters suitable as features for classifying cartilage-degradation stages. The basis for the numerical ultrasound simulations will be a novel 3D acoustical and morphological model of human hyaline cartilage. The model will allow us to test various OA-related cartilage properties independent from each other and will help us to optimize the QUS estimates derived from ex vivo QUS measurements. We will combine well established and novel QUS estimates and will develop new signal-processing approaches to compute these parameters. Nonlinear classifiers for cartilage degeneration will be developed and optimized using ROC methods, and will be tested on an existing data base of ultrasound data from histologically evaluated OA patients. If this project is successful, subsequent projects will refine the developed tools and implement them in a clinical scanner.
Currently, no available diagnostic modality is able to detect early signs of osteoarthritis reliably and thereby to allow effective treatment of the disease. We propose to develop a non-invasive, ultrasound-based classification tool for detection of early micro-morphological osteoarthritis associated with cellular changes in cartilage. The tool will be developed using a combination of ex-vivo ultrasound measurements, new backscatter ultrasound models, and numerical simulations.
