UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Feature extraction to aid disease detection and assessment of disease progression in CT and MR colonography

Hampshire, TE; (2015) Feature extraction to aid disease detection and assessment of disease progression in CT and MR colonography. Doctoral thesis , UCL (University College London). Green open access

[thumbnail of main_compress.pdf]
Preview
PDF
main_compress.pdf

Download (13MB)

Abstract

Computed tomographic colonography (CTC) is a technique employed to examine the whole colon for cancers and premalignant adenomas (polyps). Oral preparation is taken to fully cleanse the colon, and gas insufflation maximises the attenuation contrast between the enoluminal colon surface and the lumen. The procedure is performed routinely with the patient both prone and supine to redistribute gas and residue. This helps to differentiate fixed colonic pathology from mobile faecal residue and also helps discover pathology occluded by retained fluid or luminal collapse. Matching corresponding endoluminal surface locations with the patient in the prone and supine positions is therefore an essential aspect of interpretation by radiologists; however, interpretation can be difficult and time consuming due to the considerable colonic deformations that occur during repositioning. Hence, a method for automated registration has the potential to improve efficiency and diagnostic accuracy. I propose a novel method to establish correspondence between prone and supine CT colonography acquisitions automatically. The problem is first simplified by detecting haustral folds which are elongated ridgelike endoluminal structures and can be identified by curvature based measurements. These are subsequently matched using appearance based features, and their relative geometric relationships. It is shown that these matches can be used to find correspondence along the full length of the colon, but may also be used in conjunction with other registration methods to achieve a more robust and accurate result, explicitly addressing the problem of colonic collapse. The potential clinical value of this method has been assessed in an external clinical validation, and the application to follow-up CTC surveillance has been investigated. MRI has recently been applied as a tool to quantitatively evaluate the therapeutic response to therapy in patients with Crohn's disease, and is the preferred choice for repeated imaging. A primary biomarker for this evaluation is the measurement of variations of bowel wall thickness on changing from the active phase of the disease to remission; however, a poor level of interobserver agreement of measured thickness is reported and therefore a system for accurate, robust and reproducible measurements is desirable. I propose a novel method which will automatically track sections of colon, by estimating the positions of elliptical cross sections. Subsequently, estimation of the positions of the inner and outer bowel walls are made based on image gradient information and therefore a thickness measurement value can be extracted.

Type: Thesis (Doctoral)
Title: Feature extraction to aid disease detection and assessment of disease progression in CT and MR colonography
Open access status: An open access version is available from UCL Discovery
Language: English
Keywords: Registration, medical image computing, segmentation, colon, CTC, crohn's disease, Inflammatory bowel disease, virtual colonoscopy, CT colonography, MRI, markov random field, particle filter, inference, haustral folds
UCL classification: UCL > Provost and Vice Provost Offices
UCL > Provost and Vice Provost Offices > UCL BEAMS
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Med Phys and Biomedical Eng
URI: https://discovery.ucl.ac.uk/id/eprint/1461345
Downloads since deposit
239Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item