Related Experiment Video
Updated: May 3, 2026
15:26
3D Modeling of the Lateral Ventricles and Histological Characterization of Periventricular Tissue in Humans and Mouse
Published on: May 19, 2015
13.8K
Lateral ventricle segmentation of 3D pre-term neonates US using convex optimization
Wu Qiu1, Jing Yuan1, Jessica Kishimoto1
1Robarts Research Institute, University of Western Ontario, London, ON, Canada.
Medical Image Computing and Computer-Assisted Intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention
|February 8, 2014
Summary
A new 3D ultrasound segmentation method accurately measures lateral ventricle volume in preterm infants with intraventricular hemorrhage (IVH). This approach improves monitoring of ventricular dilation, offering enhanced accuracy and efficiency for neonates at risk.
Area of Science:
- Medical Imaging
- Neonatal Medicine
- Computational Imaging
Background:
- Intraventricular hemorrhage (IVH) affects 12-20% of preterm infants (<35 weeks gestational age).
- Conventional 2D ultrasound (US) for IVH monitoring has limitations in assessing ventricular dilation compared to volumetric methods.
- 3D US offers more sensitive volumetric measurements for tracking longitudinal changes in ventricular volume.
Purpose of the Study:
- To develop and evaluate a novel semi-automatic segmentation approach for lateral ventricles in preterm neonates with IVH using 3D US.
- To improve the accuracy and efficiency of ventricular volume measurement in this vulnerable population.
Main Methods:
- A global optimization-based surface evolution approach was employed for lateral ventricle segmentation.
- The method utilizes convex optimization, a subject-specific shape model, and convex relaxation to solve a combinatorial optimization problem.
- A coupled continuous max-flow model and a dual-based algorithm implemented on GPUs were developed for efficient computation.
Main Results:
- The proposed method demonstrated advantages in both accuracy and efficiency for segmenting lateral ventricles in neonates with IVH.
- The approach successfully addressed the challenges of combinatorial optimization through convex relaxation.
- High-performance numerical computation was achieved via GPU implementation.
Conclusions:
- This study presents the first semi-automatic segmentation of lateral ventricles in neonates with IVH from 3D US images.
- The developed method offers a more accurate and efficient tool for monitoring ventricular changes in preterm infants at risk of IVH.
- The findings highlight the potential of advanced computational imaging techniques in neonatal care.
