Accurate and Robust Brain Image Alignment using Boundary-based Registration

Accurate and Robust Brain Image Alignment using Boundary-based Registration

2009 October 15; 48(1): 63–72 | Douglas N. Greve and Bruce Fischl
The paper introduces a new brain image alignment algorithm called Boundary-Based Registration (BBR) to address the challenges of integrating information from different brain images, especially when there are strong intensity gradients or only partial brain images are available. BBR treats the reference image, which must be of high quality to extract surfaces separating tissue types, and the input image, which can be any modality with tissue contrast, by maximizing the intensity gradient across tissue boundaries. The algorithm is shown to be more accurate and robust than existing methods like Cross-Correlation (CR) and Normalized Mutual Information (NMI), particularly in handling intensity inhomogeneities and partial brain images. BBR's performance is evaluated through visual inspection, fMRI activation mapping, and robustness tests, demonstrating superior accuracy and robustness under various conditions, including B0 distortion, reduced field-of-view, and intensity bias. The software implementing BBR is publicly available as part of the FreeSurfer package.The paper introduces a new brain image alignment algorithm called Boundary-Based Registration (BBR) to address the challenges of integrating information from different brain images, especially when there are strong intensity gradients or only partial brain images are available. BBR treats the reference image, which must be of high quality to extract surfaces separating tissue types, and the input image, which can be any modality with tissue contrast, by maximizing the intensity gradient across tissue boundaries. The algorithm is shown to be more accurate and robust than existing methods like Cross-Correlation (CR) and Normalized Mutual Information (NMI), particularly in handling intensity inhomogeneities and partial brain images. BBR's performance is evaluated through visual inspection, fMRI activation mapping, and robustness tests, demonstrating superior accuracy and robustness under various conditions, including B0 distortion, reduced field-of-view, and intensity bias. The software implementing BBR is publicly available as part of the FreeSurfer package.
Reach us at info@study.space