Djurabekova, Nargiza; (2022) Orthopaedic dynamic imaging using ConeBeam CT. Doctoral thesis (Ph.D), UCL (University College London).

Preview	Text Nargiza_PhD_thesis.pdf - Accepted Version Download (37MB) | Preview
	Text Nargiza thesis - Supplementary materials.ppsx - Supplemental Material Download (129MB)

Abstract

With this PhD project, we investigate the feasibility of imaging the movement of a weight-bearing foot and ankle using a conebeam CT scanner. This novel approach to studying orthopaedic dynamics opens a lot of new avenues for research and our aim is to explore clinically relevant and mathematically sound dynamic reconstruction techniques. The reconstruction algorithms in this project were tested on four different phantoms of our making: two digital and two physical. We begin by considering a specialized imaging protocol consisting of two consecutive scans - one static and one dynamic. The static scan acts as the initial position of the foot and ankle and offers a wealth of information about the shape and location of the bones in a weight-bearing foot and ankle. The dynamic scan then captures movement which we assume to be reasonably periodic and traceable so that it can be binned into timeframes. We develop four different reconstruction techniques spanning from fully parametric to completely image-based. For the first one, we align individual bones from the static scan to those in the dynamic one based on their principal components. With the second approach, we register bones from a static 3D volume to 2D fluoroscopic images. Then to work on the whole foot simultaneously instead of individual bones, we move away from fully parametric approaches toward variational models. One of these models consists of joint reconstruction and motion estimation, while another also estimates segmentation. In both cases, the rigid motion of multiple bones is captured with optical flow while segmentation is performed using the Chan-Vese method to differentiate between air and bone. Both the joint methods use block gradient descent schemes known as Proximal Alternating Linearized Minimization for its global convergence and its sped-up, inertial counterpart. These methods show clear improvement over the static frame-by-frame reconstruction of binned dynamic data using FISTA, particularly with a low number of projection angles. The low number of binned projection angles for the dynamic scan ensures minimal exposure of the patient to ionising radiation.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item