Publication Details



Musculoskeletal (MSK) conditions represent a leading cause of physical disability and account for a significant proportion of lost productivity in the workplace. Understanding these conditions and improving the results of therapeutic interventions remains a huge challenge. Standard radiological imaging such as radiographs, conventional computed tomography (CT) and Magnetic Resonance Imaging MRI provide morphological and structural information but do not capture functional information on (intra-articular) motion. Obtaining combined functional and morphological information from radiological imaging is expected to further enhance diagnosis. Wide beam CT scanners allow to acquire 3D radiographic images from moving anatomical structures over lengths of up to 16 cm. This technique also referred to as 4D-CT, has the benefit of allowing the evaluation of MSK conditions which only present during dynamic activities. The technique is however not yet developed for routine clinical applications. Current, challenges include the potential radiation burden to the patient, the quality of the obtained dynamic images and the time and effort required to process the images. This PhD thesis proposes a number of contributions towards the improvement and usability of 4D-CT imaging for real-time motion of MSK structures. In chapter 1, we provide background information about the topic and pave the way for the technical contributions presented in the subsequent chapters. In chapter 2, we present a feasibility study to investigate the ability of 4D-CT to detect progressive ligament cuts in a cadaver foot using a low radiation scan protocol. Results demonstrated the potential to detect differences in kinematics due to changes in ligamentintegrity. Motion artifacts were however still present even at the fastest tube rotation, warranting the need for further scan protocol optimization. In chapter 3, we investigated the potential of tracking mandibular joint motion using 4D CT. Complex mandibular motions were simulated using a mandibular phantom, mounted on a six degrees of freedom simulator. Plots of mandibular range of motion and 3Dcondyle trajectories were obtained from these experiments with reproducible results. Future applications could be in pre-operative planning or potential patient follow-up from c ombining the visual and 3D graphical mandibular motion kinematics. In chapter 4, we addressed the issue of motion artifacts by leveraging the significant advances in cardiac imaging applications to extend a cardiac acquisition protocol to dynamic MSK applications. We designed a rotating phantom to quantitatively assess thevalue of the cardiac acquisition mode over a cine dynamic mode in terms motion artifacts.Our findings demonstrate that dynamic MSK images can successfully be obtained using the cardiac scan mode, as it significantly reduces motion artifacts with respect to cine acquisitions. In chapter 5, an automated image analysis workflow for processing dynamic MSK im ages is presented. A multi-atlas, multi-label segmentation and landmark propagation approach is proposed for performing automated kinematic analysis of dynamic CT MSK images. The workflow was evaluated on MSK structures of the thumb and knee and was found to perform on par with manual analysis performed by experts. In chapter 6, we apply the knowledge and tools from the previous chapters to investigate patellar motion of healthy subjects in weight-bearing conditions. We introduce a novel weight bearing device that enables the simulation of an orthostatic squat during dynamic images in a supine position. Our weight-bearing device combined with our automated workflow successfully provide normative values of healthy patellar motion. This tech nique could potentially serve as an objective tool for the functional evaluation of patella tracking. Finally in chapter 7, we investigated the thumb joint of healthy subjects during cyclic opposition reposition motion with our automated workflow. Cardan angles of the first metacarpal (MC1) and trapezium (Tz) were obtained relative to the second metacarpal MC2 and presented in function of time. The area of the articulating surfaces of MC1 and Tz were estimated as a surrogate for joint congruence and presented as color-coded joint proximity maps in function of the motion of the joint. We observed motion-dependent proximity patterns which could be used in future as a model to investigate early onset of Osteoarthrosis (OA). In conclusion, the contributions of this thesis provide a significant time gain in process ing the huge amount of images generated from 4D-CT MSK data. It helps minimise the laborious post-processing tasks by reducing user-interaction steps, primarily in bone seg mentations and anatomical landmark placement. We have demonstrated the possibility of significantly reducing motion artifact by leveraging a cardiac acquisition protocol for acquiring 4D-CT MSK images at reasonable radiation burden to the patient. These contributions offer a framework for future clinical research, to improve our under standing of joint instabilities and disorders. When successfully integrated into clinical workflow, it could benefit individual patient care through early diagnosis and treatment of certain MSK pathologies. It could equally be used for patient follow-up and potentially inform clinical decisions