Optimizing Diffusion MRI Strategies: Gradient Sampling Schemes and b-Values for Infants

1599 

Abstract Submission 

Yajuan Zhang¹, Rui Zhou¹, Lin Zhang², Mingwen Yang², Zuozhen Lan², Ying Lin³, Lixuan Zhu¹, Guoqiang Cheng⁴, Xianghui Huang³, Jungang Liu², Han Zhang^1,5

¹School of Biomedical Engineering, ShanghaiTech University, Shanghai, China, ²Department of Radiology, Xiamen Children's Hospital, Children's Hospital of Fudan University, Xiamen, Fujian, China, ³Fujian Key Laboratory of Neonatal Diseases, Children’s Hospital of Fudan University (Xiamen Branch), Xiamen, Fujian, China, ⁴Department of Neonatology,Children’s Hospital of FudanUniversity, National Children’s Medical Center, Shanghai, China, ⁵Shanghai Clinical Research and Trial Center, Shanghai, China

Yajuan Zhang  
School of Biomedical Engineering, ShanghaiTech University
Shanghai, China

Rui Zhou  
School of Biomedical Engineering, ShanghaiTech University
Shanghai, China

Lin Zhang  
Department of Radiology, Xiamen Children's Hospital, Children's Hospital of Fudan University
Xiamen, Fujian, China

Mingwen Yang  
Department of Radiology, Xiamen Children's Hospital, Children's Hospital of Fudan University
Xiamen, Fujian, China

Zuozhen Lan  
Department of Radiology, Xiamen Children's Hospital, Children's Hospital of Fudan University
Xiamen, Fujian, China

Ying Lin  
Fujian Key Laboratory of Neonatal Diseases, Children’s Hospital of Fudan University (Xiamen Branch)
Xiamen, Fujian, China

Lixuan Zhu  
School of Biomedical Engineering, ShanghaiTech University
Shanghai, China

Guoqiang Cheng  
Department of Neonatology,Children’s Hospital of FudanUniversity, National Children’s Medical Center
Shanghai, China

Xianghui Huang  
Fujian Key Laboratory of Neonatal Diseases, Children’s Hospital of Fudan University (Xiamen Branch)
Xiamen, Fujian, China

Jungang Liu  
Department of Radiology, Xiamen Children's Hospital, Children's Hospital of Fudan University
Xiamen, Fujian, China

Han Zhang  
School of Biomedical Engineering, ShanghaiTech University|Shanghai Clinical Research and Trial Center
Shanghai, China|Shanghai, China

Multi-shell dMRI with simultaneous multi-slice imaging technique allows for characterizing complex diffusion signals with the help of advanced models such as neurite orientation dispersion and density imaging (NODDI) (Zhang et al., 2012) and constrained spherical deconvolution (CSD) (Jeurissen et al., 2014). However, the significantly prolonged acquisition time poses potential challenges, especially for infant where the available acquisition time for non-sedated healthy infants is limited. Furthermore, the infant brain consists of more water and less myelin; adult protocol cannot be directly applied. For large-scale infant brain development cohort (age 0-6 years), it is very critical and necessary to develop and valid a dMRI protocol balancing image quality, scanning time, and the dramatically changed macro-/micro-structure and connectomic attributes during early development. Taken together, this study did not only refer the existing infant cohorts (e.g., dHCP and BCP) (Bastiani et al., 2019; Howell et al., 2019) but also proposed modifications to the gradient sampling protocol and b-values, aiming to establishing infant dMRI protocol that offers accurate modeling while maintaining flexibility, quality, success rate across 0-6 years of age.

The infant dMRI protocol optimization is carried out for the China Baby Connectome Project (CBCP) as a pilot study. In this study, 5 subjects were recruited at 1 wk, 1 mo, 2 mo, 3 yrs, and 5 yrs, respectively. Multi-shell dMRI (TR/TE=3016/77.1ms, FOV=210×210 mm2, slices number=92, slice thickness=1.5 mm, multiband factor=4) were acquired with a 3.0T scanner (uMR890, United Imaging). To assess various candidate multi-shell schemes and establish an optimal one for infant dMRI longitudinal cohort, two experiments were conducted. Experiment 1: Comparing two gradient encoding schemes: 1) complete acquisitions of three-shell scheme (500, 1000, 3000 s/mm²) with 9, 12, and 48 directions, respectively, at both AP and PA phase encoding (PE) directions (8min24s); 2) single acquisition of six b-value shells (500, 1000, 1500, 2000, 2500, 3000 s/mm²) with 9, 12, 17, 24, 34, and 48 directions, respectively, with PA PE directions, and b=0 for AP PE (8min44s). Experiment 2: Determine best b-value combinations for infants using three sampling schemes (blow=500 s/mm², bmid=1000 s/mm², bhigh=2000, or 2500, or 3000 s/mm²) with the same b-vectors. We systematically compared the derived DTI and NODDI indices and CSD-based orientation distribution function (ODF) for an optimized protocol.

In preschoolers (3 and 5 yrs old), Schemes 1 and 2 generated very similar results. In three infants (1 wk, 1 mo, 2 mo old) with immature myelination, only the main fibers were detectable by both schemes. Nevertheless, compared to Scheme 2, Scheme 1 provided more accurate fiber orientations and enhanced sensitivity to crossing fibers (Fig 1). Compared to the schemes with bhigh of 2000 and 2500 s/mm², the scheme with bhigh of 3000 s/mm² exhibited greater sensitivity in ODF estimation to crossing fibers, while maintaining comparable performance using DTI and NODDI (Fig 2).

This study established the optimal multi-shell infant dMRI acquisition scheme: three shells (500, 1000, 3000 s/mm²) with 9, 12, and 48 directions, respectively, and a complete acquisition with both AP and PA PE directions. Additionally, all b-vectors and b-shells should be optimally ordered to ensure incremental acquisition within each shell and across all the shells. A blow of 500 s/mm² ensures compatibility to low myelination and high-water content in neonates, a bhigh of 3000 s/mm² enhances sensitivity to microstructures and the crossing fibers. Combining information from both full AP and PA PE acquisitions with 3 shells improves the quality of the infant dMRI derivatives without significantly prolonging acquisition time or local distortion. Our optimized dMRI provides suitable imaging protocol for infant studies with balanced data quality and acquisition time.

Normal Brain Development: Fetus to Adolescence ²

Diffusion MRI Modeling and Analysis ¹

White Matter Anatomy, Fiber Pathways and Connectivity

Diffusion MRI

Computational Neuroscience

Experimental Design

WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC