An Evaluation of Deformation-Based Morphometry Applied to the Developing Human Brain and Detection of Volumetric Changes Associated with Preterm Birth

James P. Boardman, Kanwal K. Bhatia, Serena J. Counsell, Joanna M. Allsop, Olga Kapellou, Mary A. Rutherford, A. David Edwards, Joseph V. Hajnal, Daniel RueckertMICCAI 2003 · DOI

Abstract

Preterm birth is a significant risk factor for neurodevelopmental impairment, yet identifying the subtle structural brain changes involved remains a challenge. This study utilizes high-dimensional deformation-based morphometry (DBM) to automatically quantify volumetric differences between preterm infants and full-term controls at term-equivalent age. Our findings reveal significant tissue contraction in the deep grey matter and expansion in the ventricles of preterm infants. Crucially, we demonstrate that these morphometric results are robust and independent of the specific reference atlas used, validating DBM as a reliable tool for automated phenotyping of the developing brain.

Introduction

The rapid growth and complex maturation of the infant brain make traditional neuroimaging analysis difficult. Manual segmentation is labor-intensive and prone to observer bias, while low-dimensional alignment techniques often fail to capture fine-grained structural variations. Furthermore, automated methods frequently rely on a single "reference brain" or atlas, raising concerns about "atlas bias" where the choice of template might influence the results. This study addresses these challenges by applying high-dimensional DBM to characterize the neuroanatomical phenotype of preterm birth and evaluating the consistency of these findings across multiple reference spaces.

Methods

The study analyzed T1-weighted MR images from 66 preterm infants (scanned at term-equivalent age) and 11 term-born controls. The core methodology involved Deformation-Based Morphometry (DBM) using a non-rigid registration algorithm based on Free-Form Deformations (FFD) and B-splines.

To quantify local volume changes, the Jacobian determinant of the deformation field was calculated at every voxel. A Jacobian value greater than 1 indicates local expansion relative to the reference, while a value less than 1 indicates contraction. To test for atlas bias, the entire analysis pipeline was repeated using three different term-born infant brains as reference templates ($R_{\#1}, R_{\#2}, R_{\#3}$).

Results and Key Insights

Atlas Independence

A primary finding of this work is that DBM results in the developing brain are largely independent of the reference space. The regions identified as having significant volumetric differences remained consistent regardless of which of the three templates was employed. This suggests that high-dimensional DBM is a robust method for comparing infant brain structures without the confounding factor of template selection.

Volumetric Signatures of Preterm Birth

The analysis identified a distinct "volumetric fingerprint" in preterm infants compared to term controls:

  • Deep Grey Matter Contraction: Unexpectedly, significant volume reduction was found in the basal ganglia (globus pallidus and thalamus). In infants born before 30 weeks, these regions were reduced to approximately 72-79% of the volume seen in term controls.
  • Ventriculomegaly: Preterm infants exhibited marked expansion of the lateral ventricles, with volumes reaching 122% to 131% of the control group.
  • Extracerebral Space Expansion: Widening was observed in the interhemispheric fissure and cerebrospinal fluid spaces around the frontal lobes.
Figure 1: Localization of significant volumetric differences in preterm infants. Rows show regions of contraction in the basal ganglia (a-c), expansion in the ventricles (d-f), and expansion in the extracerebral space (g-i).

The robustness of these findings is further evidenced by the high effect sizes recorded across different reference spaces, as shown in the table below.

Table 1: Quantitative summary of effect sizes and volume changes in the basal ganglia and ventricles for different preterm subgroups across three reference spaces.

Discussion

The discovery of localized contraction in the deep grey matter is particularly noteworthy. This vulnerability might stem from direct environmental stressors in the neonatal intensive care unit or secondary effects related to altered myelination patterns in adjacent white matter tracts. While this study provides a powerful automated method for identifying structural differences, it is limited by a relatively small control group and the potential confounding effect of varying myelination on MR signal intensities. Future research should focus on longitudinal studies to correlate these early morphometric markers with long-term functional and cognitive outcomes in children born preterm.

Summary updated

BibTeX
@inproceedings{boardman2003evaluation,
  author = {James P. Boardman and Kanwal K. Bhatia and Serena J. Counsell and Joanna M. Allsop and Olga Kapellou and Mary A. Rutherford and A. David Edwards and Joseph V. Hajnal and Daniel Rueckert},
  title = {{An Evaluation of Deformation-Based Morphometry Applied to the Developing Human Brain and Detection of Volumetric Changes Associated with Preterm Birth}},
  booktitle = {International Conference on Medical Image Computing and Computer-Assisted Intervention},
  year = {2003},
  doi = {10.1007/978-3-540-39899-8\_85},
}
Figure