Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.617914
Title: Registration of prone and supine CT colonography images and its clinical application
Author: Roth, H. R.
Awarding Body: University College London (University of London)
Current Institution: University College London (University of London)
Date of Award: 2013
Availability of Full Text:
Access through EThOS:
Full text unavailable from EThOS. Please try the link below.
Access through Institution:
Abstract:
Computed tomographic (CT) colonography is a technique for detecting bowel cancer and potentially precancerous polyps. CT imaging is performed on the cleansed and insufflated bowel in order to produce a virtual endoluminal representation similar to optical colonoscopy. Because fluids and stool can mimic pathology, images are acquired with the patient in both prone and supine positions. Radiologists then match endoluminal locations visually between the two acquisitions in order to determine whether pathology is real or not. This process is hindered by the fact that the colon can undergo considerable deformation between acquisitions. Robust and accurate automated registration between prone and supine data acquisitions is therefore pivotal for medical interpretation, but a challenging problem. The method proposed in this thesis reduces the complexity of the registration task of aligning the prone and supine CT colonography acquisitions. This is done by utilising cylindrical representations of the colonic surface which reflect the colon's specific anatomy. Automated alignment in the cylindrical domain is achieved by non-rigid image registration using surface curvatures, applicable even when cases exhibit local luminal collapses. It is furthermore shown that landmark matches for initialisation improve the registration's accuracy and robustness. Additional performance improvements are achieved by symmetric and inverse-consistent registration and iteratively deforming the surface in order to compensate for differences in distension and bowel preparation. Manually identified reference points in human data and fiducial markers in a porcine phantom are used to validate the registration accuracy. The potential clinical impact of the method has been evaluated using data that reflects clinical practise. Furthermore, correspondence between follow-up CT colonography acquisitions is established in order to facilitate the clinical need to investigate polyp growth over time. Accurate registration has the potential to both improve the diagnostic process and decrease the radiologist's interpretation time. Furthermore, its result could be integrated into algorithms for improved computer-aided detection of colonic polyps.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.617914  DOI: Not available
Share: