Registration of Temporally Separated CT Colonography Cases

• Published in: Abdominal Imaging. Computation and Clinical Applications pp. 46 - 52
• Main Authors: Roth, Holger R., Helbren, Emma, Hampshire, Thomas E., Pickhardt, Perry J., Halligan, Steve, Hawkes, David J.
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2013
• Series: Lecture Notes in Computer Science
• Subjects: Abdominal imaging; computed tomography; computer-aided diagnosis; CT colonography; follow-up investigations; oncology applications; registration
• ISBN: 9783642410826; 3642410820
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-642-41083-3_6

Robust registration between prone and supine data acquisitions for CT colonography (CTC) is a useful tool for assessing clinically significant changes but a challenging problem. This is especially the case for polyp follow-up when scans are temporally separated. We investigated the ability of automatic registration to align CTC cases, acquired several months apart. 26 initial and follow-up cases were investigated and registration measured using the locations of 35 polyps in all available scans. Robust non-rigid feature-based initialization allowed registration of prone and supine CTC scans from patient cases not only acquired on the same day but also when acquired several months apart. A mean registration error of 17.4 (std. dev. 12.1) mm (median 14.9 mm, range 1.7 to 49.7 mm) was achieved when transforming polyp locations between longitudinal scans. The level of accuracy achieved was similar to previous studies that aligned CTC images acquired at the same sitting. Automatic registration of follow-up CTC investigations could be a useful adjunct for radiologists interpreting CTC for surveillance of colonic polyps.