Radiomic and Image-Based Biomechanical Approaches for Sonographic Evaluation of the Thoracolumbar Fascia in Patients With Mechanical Low Back Pain

Abstract

Low back pain (LBP) is the most common musculoskeletal ailment worldwide and a leading cause of disability. It remains highly challenging to treat clinically due to its complex, multifactorial nature. A growing body of literature provides evidence that the thoracolumbar fascia (TLF)—a densely woven network of collagen fibers that supports and stabilizes the back and trunk—contributes to LBP. Yet substantial gaps remain in our understanding of the TLF's contribution to LBP and in rigorous methods to investigate the structure and function of this tissue in vivo. Ultrasound (US) imaging holds significant promise for point-of-care clinical assessment of the TLF, however there exists limited evidence that commonly used morphological assessments, such as thickness and echogenicity, can reliably distinguish the TLF of individuals with and without LBP. There is a critical need for reliable US-based methods of evaluating the TLF to aid in diagnosis, treatment, and monitoring of LBP. Radiomics, an emerging technique that uses mathematical algorithms to extract quantitative features from medical images, presents a promising approach to augment conventional US image interpretation methods. Image-based biomechanical techniques to quantify movement and tissue properties from medical images and videos can provide further insight into the tissue characteristics and behavior. The work presented in this dissertation focuses on the development and evaluation of rigorous, sonography-based methods of evaluating the TLF to support diagnosis, treatment and monitoring of TLF-associated LBP. The techniques investigated include automated segmentation to aid in distinguishing the TLF within sonograms, sonogram standardization methods to address variations arising from differing image acquisition conditions, methods of estimating TLF displacement from ultrasound recordings, and the utility of radiomic techniques for automating clinical interpretation of TLF sonograms. Following development and evaluation of these radiomic and image-based biomechanical techniques, they were applied in conjunction with multimodal US imaging to compare TLF characteristics and function in individuals with and without LBP.