Segmenting hypothalamic subunits in human newborn magnetic resonance imaging data

Abstract

Preclinical evidence suggests that inter-individual variation in the structure of the hypothalamus at birth is associated with variation in the intrauterine environment, with downstream implications for future disease susceptibility. However, scientific advancement in humans is limited by a lack of validated methods for the automatic segmentation of the newborn hypothalamus. N = 215 healthy full-term infants with paired T1-/T2-weighted MR images across four sites were considered for primary analyses (mean postmenstrual age = 44.3 ± 3.5 weeks, n_male /n_female = 110/106). The outputs of FreeSurfer's hypothalamic subunit segmentation tools designed for adults (segFS) were compared against those of a novel registration-based pipeline developed here (segATLAS) and against manually edited segmentations (segMAN) as reference. Comparisons were made using Dice Similarity Coefficients (DSCs) and through expected associations with postmenstrual age at scan. In addition, we aimed to demonstrate the validity of the segATLAS pipeline by testing for the stability of inter-individual variation in hypothalamic volume across the first year of life (n = 41 longitudinal datasets available). SegFS and segATLAS segmentations demonstrated a wide spread in agreement (mean DSC = 0.65 ± 0.14 SD; range = {0.03-0.80}). SegATLAS volumes were more highly correlated with postmenstrual age at scan than segFS volumes (n = 215 infants; R_segATLAS ² = 65% vs. R_segFS ² = 40%), and segATLAS volumes demonstrated a higher degree of agreement with segMAN reference segmentations at the whole hypothalamus (segATLAS DSC = 0.89 ± 0.06 SD; segFS DSC = 0.68 ± 0.14 SD) and subunit levels (segATLAS DSC = 0.80 ± 0.16 SD; segFS DSC = 0.40 ± 0.26 SD). In addition, segATLAS (but not segFS) volumes demonstrated stability from near birth to ~1 years age (n = 41; R² = 25%; p < 10^-3 ). These findings highlight segATLAS as a valid and publicly available (https://github.com/jerodras/neonate_hypothalamus_seg) pipeline for the segmentation of hypothalamic subunits using human newborn MRI up to 3 months of age collected at resolutions on the order of 1 mm isotropic. Because the hypothalamus is traditionally understudied due to a lack of high-quality segmentation tools during the early life period, and because the hypothalamus is of high biological relevance to human growth and development, this tool may stimulate developmental and clinical research by providing new insight into the unique role of the hypothalamus and its subunits in shaping trajectories of early life health and disease.

Keywords: MRI; growth; hypothalamus; infant; newborn; segmentation; subunit.

FIGURE 1

Atlas‐based registration pipeline (segATLAS). A registration pipeline was designed to leverage high quality age‐appropriate atlases in combination with a robust brain extraction procedure. Collectively, the proposed pipeline uses an age‐appropriate middle registration point to warp a FreeSurfer compatible definition of hypothalamic subunits into native infant space for segmentation. BCP, Baby Connectome Project; MNI, Montreal Neurological Institute.

FIGURE 2

Summary of comparison between Naïve FreeSurfer (segFS) and atlas‐based registration (segATLAS) approaches to segmentation. Segmentation approaches (segFS and segATLAS) varied in their degree of agreement (example of good agreement, a). SegFS had several instances of gross underestimation (e.g., data points below 200mm³, b), and segATLAS volume was more strongly associated with postmenstrual age at scan than segFS (b). Segmentations in disagreement (segFS vs. segATLAS) tended to be younger (c) and underestimated by segFS (d, white border indicates whole‐hypothalamus border in template space).

FIGURE 3

Comparison between manual corrections (segMAN) and automatic approaches to segmentation (segFS and segATLAS). Segmentation approaches were compared against manually edited standards. Example segmentations are shown in the top row (note the fornix is properly excluded from the segmentation, white arrow). Bottom left depicts whole hypothalamic volume across methods. SegFS volumes had a relatively widespread encompassing some implausibly small volumes. SegMAN volumes were smaller on average as they were biased (the intersection of segATLAS and segMAS as the initialization point). Bottom right depicts the DSC values from each of the three segmentation approaches in reference to segMAN segmentations.