Evaluation of Five Diffeomorphic Image Registration Algorithms for Mouse Brain Magnetic Resonance Microscopy.

Zhenrong Fu,Lan Lin,Miao Tian,Jingxuan Wang,Baiwen Zhang,Pingping Chu,Shaowu Li,Muhammad Mohsin Pathan,Yulin Deng,Shuicai Wu
DOI: https://doi.org/10.1111/jmi.12594
IF: 1.9522
2017-01-01
Journal of Microscopy
Abstract:The development of genetically engineered mouse models for neuronal diseases and behavioural disorders have generated a growing need for small animal imaging. High-resolution magnetic resonance microscopy (MRM) provides powerful capabilities for noninvasive studies of mouse brains, while avoiding some limits associated with the histological procedures. Quantitative comparison of structural images is a critical step in brain imaging analysis, which highly relies on the performance of image registration techniques. Nowadays, there is a mushrooming growth of human brain registration algorithms, while fine-tuning of those algorithms for mouse brain MRMs is rarely addressed. Because of their topology preservation property and outstanding performance in human studies, diffeomorphic transformations have become popular in computational anatomy. In this study, we specially tuned five diffeomorphic image registration algorithms [DARTEL, geodesic shooting, diffeo-demons, SyN (Greedy-SyN and geodesic-SyN)] for mouse brain MRMs and evaluated their performance using three measures [volume overlap percentage (VOP), residual intensity error (RIE) and surface concordance ratio (SCR)]. Geodesic-SyN performed significantly better than the other methods according to all three different measures. These findings are important for the studies on structural brain changes that may occur in wild-type and transgenic mouse brains.
What problem does this paper attempt to address?