Xianjun Li1, Jie Gao2, Miaomiao Wang2, Mingxi Wan3, Jian Yang4. 1. Department of Biomedical Engineering, The Key Laboratory of Biomedical Information Engineering of the Ministry of Education, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, China; Department of Radiology, The First Affiliated Hospital of Medical School, Xi'an Jiaotong University, Xi'an, Shaanxi, China. 2. Department of Radiology, The First Affiliated Hospital of Medical School, Xi'an Jiaotong University, Xi'an, Shaanxi, China. 3. Department of Biomedical Engineering, The Key Laboratory of Biomedical Information Engineering of the Ministry of Education, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, China. Electronic address: mxwan@mail.xjtu.edu.cn. 4. Department of Biomedical Engineering, The Key Laboratory of Biomedical Information Engineering of the Ministry of Education, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, China; Department of Radiology, The First Affiliated Hospital of Medical School, Xi'an Jiaotong University, Xi'an, Shaanxi, China. Electronic address: cjr.yangjian@vip.163.com.
Abstract
PURPOSE: The relatively poor image contrast and variation in the neonatal brain size are technical challenges associated with the typical tract-based spatial statistics (TBSS) for the target identification and normalization. This study aimed to develop a rapid and reliable pipeline for the neonatal TBSS. MATERIALS AND METHODS: A rapid TBSS strategy was proposed based on the group-wise target choice for fractional anisotropy (FA) derived from diffusion tensor imaging (DTI). The most representative subject of the entire group was identified via (a) initial group-averaged template creation (b) followed by identification of the target with the minimum warp displacement score between the individual and the group-averaged template. The computation time, registration quality, measurement of regional values, and statistical analyses were evaluated in two applications: brain white matter development in normal term neonates, and alterations in preterm neonates with white matter lesions compared to the matched controls. These performances in the proposed pipeline were compared with those in the typical and previous neonatal TBSS workflows. RESULTS: Target choice using the proposed strategy is faster, compared with the previous TBSS pipelines, especially with the increase of the sample size. Registration errors between individuals and the target are assessed through warp displacement scores. Smaller warp displacement scores are observed for the proposed method than the typical pipeline. Due to the relatively accurate registration, the proposed method results in lower standard deviations and higher averaged values of FA across subjects. Additionally, more areas with significant changes related to the development and white matter lesions are detected using the proposed method than previous TBSS pipelines. The proposed pipeline provides stronger correlation between FA and gestational age, and larger difference between preterm neonates with white matter lesions and controls. CONCLUSION: The proposed TBSS pipeline improves the efficiency and reliability of the DTI analysis in the neonatal brain.
PURPOSE: The relatively poor image contrast and variation in the neonatal brain size are technical challenges associated with the typical tract-based spatial statistics (TBSS) for the target identification and normalization. This study aimed to develop a rapid and reliable pipeline for the neonatal TBSS. MATERIALS AND METHODS: A rapid TBSS strategy was proposed based on the group-wise target choice for fractional anisotropy (FA) derived from diffusion tensor imaging (DTI). The most representative subject of the entire group was identified via (a) initial group-averaged template creation (b) followed by identification of the target with the minimum warp displacement score between the individual and the group-averaged template. The computation time, registration quality, measurement of regional values, and statistical analyses were evaluated in two applications: brain white matter development in normal term neonates, and alterations in preterm neonates with white matter lesions compared to the matched controls. These performances in the proposed pipeline were compared with those in the typical and previous neonatal TBSS workflows. RESULTS: Target choice using the proposed strategy is faster, compared with the previous TBSS pipelines, especially with the increase of the sample size. Registration errors between individuals and the target are assessed through warp displacement scores. Smaller warp displacement scores are observed for the proposed method than the typical pipeline. Due to the relatively accurate registration, the proposed method results in lower standard deviations and higher averaged values of FA across subjects. Additionally, more areas with significant changes related to the development and white matter lesions are detected using the proposed method than previous TBSS pipelines. The proposed pipeline provides stronger correlation between FA and gestational age, and larger difference between preterm neonates with white matter lesions and controls. CONCLUSION: The proposed TBSS pipeline improves the efficiency and reliability of the DTI analysis in the neonatal brain.