α-Information-Based Registration of Dynamic Scans for Magnetic Resonance Cystography

• Published in: IEEE journal of biomedical and health informatics Vol. 20; no. 4; pp. 1160 - 1170
• Main Authors: Han, Hao, Lin, Qin, Li, Lihong, Duan, Chaijie, Lu, Hongbing, Li, Haifang, Yan, Zengmin, Fitzgerald, John, Liang, Zhengrong
• Format: Journal Article
• Language: English
• Published: United States IEEE 01-07-2016
• Subjects: Aged; Algorithms; Biomedical measurement; Bladder; Bladder cancer; Cancer; cystography; Cystography - methods; Female; Humans; Image registration; Image segmentation; Imaging, Three-Dimensional - methods; magnetic resonance (MR); Magnetic Resonance Imaging - methods; Male; Middle Aged; Phantoms, Imaging; Signal to noise ratio; Three-dimensional displays; Urinary Bladder - diagnostic imaging; Urinary Bladder Neoplasms - diagnostic imaging
• ISSN: 2168-2194; 2168-2208
• DOI: 10.1109/JBHI.2015.2441744

To continue our effort on developing magnetic resonance (MR) cystography, we introduce a novel nonrigid 3-D registration method to compensate for bladder wall motion and deformation in dynamic MR scans, which are impaired by relatively low signal-to-noise ratio in each time frame. The registration method is developed on the similarity measure of α-information, which has the potential of achieving higher registration accuracy than the commonly used mutual information (MI) measure for either monomodality or multimodality image registration. The α-information metric was also demonstrated to be superior to both the mean squares and the cross-correlation metrics in multimodality scenarios. The proposed α-registration method was applied for bladder motion compensation via real patient studies, and its effect to the automatic and accurate segmentation of bladder wall was also evaluated. Compared with the prevailing MI-based image registration approach, the presented α-information-based registration was more effective to capture the bladder wall motion and deformation, which ensured the success of the following bladder wall segmentation to achieve the goal of evaluating the entire bladder wall for detection and diagnosis of abnormality.