Shape Characterization of the Corpus Callosum in Schizophrenia Using Template Deformation

Abraham Dubb, Brian B. Avants, Ruben C. Gur, James C. GeeMICCAI 2002 · DOI

Introduction

Schizophrenia is increasingly conceptualized as a "disconnection syndrome," where the primary pathology lies in the communication pathways between brain regions. The corpus callosum (CC), the brain's largest white-matter bridge, is a logical focal point for neuroimaging research. However, decades of study have yielded inconsistent results regarding its size and shape in patients. Researchers at the University of Pennsylvania addressed these inconsistencies by applying Template Deformation Morphometry (TDM) to map localized shape differences in a large cohort of 100 schizophrenia patients and 190 healthy controls.

Methods

Traditional morphometric studies often rely on coarse metrics like total area, which can mask subtle, localized variations. To overcome this, the authors developed a pipeline centered on TDM.

First, 3D MRI volumes were manually aligned to the mid-sagittal plane using anatomical landmarks such as the falx cerebri and the anterior/posterior commissures to ensure consistent reslicing. Following alignment, the CC was isolated using a $k$-means clustering algorithm. To ensure robustness against variations in image brightness, the team generated 18 different segmentations per image and selected the most anatomically accurate result.

MRI alignment and segmentation process showing anatomical landmarks and the resulting k-means clustering.

The core of the analysis involved curve registration. A dynamic programming approach established a 1D correspondence between a "template" CC boundary and each subject's CC. This mapping was then interpolated into the structure's interior using Laplace’s equation. Finally, the Jacobian determinant of the resulting deformation field was calculated to quantify the local stretching or shrinking required to match the template, providing a normalized measure of pure shape variation.

Experimental Setup

The study utilized a dataset of 290 cranial MRIs from the Schizophrenia Center at the University of Pennsylvania. The cohort included:

  • Healthy Controls: 190 individuals (average age 28).
  • Schizophrenia Patients: 100 individuals (average age 33.7).

All participants were right-handed. Scans were acquired with 1mm slice thickness on a 1.5T GE scanner.

Results and Discussion

The TDM analysis revealed that schizophrenia is characterized by focal, rather than global, changes in the CC.

Focal Anterior Contraction

The most significant finding was a substantial area of contraction in the anterior third of the CC. This region connects the frontal lobes, which are critical for executive function—a domain frequently impaired in schizophrenia. This contraction was notably more pronounced in female patients. In contrast, male patients exhibited a small area of expansion in the mid-body of the CC.

Statistical maps showing regions of significant shape variation (contraction/expansion) across the corpus callosum template.

Developmental Stagnation

The study also identified a disruption in normal brain maturation. In healthy controls, the posterior region of the CC (the splenium) typically expands as individuals age from their 20s into their 50s. However, this age-related white-matter expansion was virtually absent in the patient group, suggesting a failure of late-stage neurodevelopmental processes.

Conclusion

By moving beyond simple area measurements to high-resolution deformation maps, this study provides a more nuanced view of callosal pathology in schizophrenia. The findings of anterior contraction and splenial stagnation offer specific anatomical targets for understanding the "disconnection" underlying the disorder.

While these results conflict with some smaller studies that reported anterior thickening, the authors suggest that sample size and registration methodology are critical factors. Future research will focus on longitudinal tracking to confirm these developmental trends and extending TDM to full 3D white-matter volumes.

Summary updated

BibTeX
@inproceedings{dubb2002shape,
  author = {Abraham Dubb and Brian B. Avants and Ruben C. Gur and James C. Gee},
  title = {{Shape Characterization of the Corpus Callosum in Schizophrenia Using Template Deformation}},
  booktitle = {International Conference on Medical Image Computing and Computer-Assisted Intervention},
  year = {2002},
  doi = {10.1007/3-540-45787-9\_48},
}
Figure