Optimal setting of image bounding box can improve registration accuracy of diffusion tensor tractography

• Published in: International journal for computer assisted radiology and surgery Vol. 9; no. 2; pp. 333 - 339
• Main Authors: Yoshino, Masanori, Kin, Taichi, Saito, Toki, Nakagawa, Daichi, Nakatomi, Hirofumi, Kunimatsu, Akira, Oyama, Hiroshi, Saito, Nobuhito
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.03.2014
• Subjects: Computer Imaging; Computer Science; Health Informatics; Imaging; Medicine; Medicine & Public Health; Original Article; Pattern Recognition and Graphics; Radiology; Surgery; Vision; Normalized mutual information method; acoustic tumor; Posterior fossa; Echo planar imaging; Bounding box; Registration
• ISSN: 1861-6410; 1861-6429
• DOI: 10.1007/s11548-013-0934-3

Purpose When we register diffusion tensor tractography (DTT) to anatomical images such as fast imaging employing steady-state acquisition (FIESTA), we register the B0 image to FIESTA. Precise registration of the DTT B0 image to FIESTA is possible with non-rigid registration compared to rigid registration, although the non-rigid methods lack convenience. We report the effect of image data bounding box settings on registration accuracy using a normalized mutual information (NMI) method Methods MRI scans of 10 patients were used in this study. Registration was performed without modification of the bounding box in the control group, and the results were compared with groups re-registered using multiple bounding boxes limited to the region of interest (ROI). The distance of misalignment after registration at 3 anatomical characteristic points that are common to both FIESTA and B0 images was used as an index of accuracy. Results Mean ( ± SD) misalignment at the 3 anatomical points decreased significantly from 5.99 ± 1.58 to 2.21 ± 1.24  mm, p < 0.0001 ), 4.36 ± 1.58 to 1.48 ± 0.58  mm, ( p < 0.0001 ) , and 5.21 ± 1.76 to 1.20 ± 0.74  mm, ( p < 0.0001 ) , each showing improvement compared to the control group Conclusion Narrowing the image data bounding box to the ROI improves the accuracy of registering B0 images to FIESTA by NMI method. With our proposed methodology, accuracy can be improved in extremely easy steps, and this methodology may prove useful for DTT registration to anatomical image