Related Experiment Video
Updated: Apr 5, 2026
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
Automatic lateral ventricle and choroid plexus segmentation method in infant brain MR images
Junghwa Kang1, Hyun Gi Kim2,3, Na-Young Shin4,5
1Department of Biomedical Engineering, Hankuk University of Foreign Studies, Yongin-si, Gyeonggi-do, 17035, Korea.
Insights
A new deep learning method accurately segments lateral ventricles (LV) and choroid plexus (CP) in infant brain MRIs. This automated approach ensures anatomical consistency, aiding studies on early neurodevelopment.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Deep Learning
Background:
- Accurate segmentation of lateral ventricles (LV) and choroid plexus (CP) in infant brain MRI is crucial for understanding cerebrospinal fluid dynamics and early neurodevelopment.
- Existing segmentation methods for adult populations often perform poorly on infant data due to anatomical differences, low contrast, and motion artifacts.
- These methods frequently result in misclassification of tissue boundaries in pediatric neuroimaging.
Purpose of the Study:
- To develop a fully automated deep learning method for joint segmentation of LV and CP in infant brain MRI.
- To address the challenges of rapid anatomical changes, low tissue contrast, and motion artifacts in infant neuroimaging.
- To validate the clinical adaptability and performance of the proposed segmentation method on independent datasets.
Main Methods:
- A fully automated deep learning approach for joint LV and CP segmentation using T1-weighted MRI was developed.
- An anatomy-aware loss function was integrated to enforce the topological constraint of CP containment within the LV.
- The method was validated on two independent datasets (Baby Connectome Project and an in-house dataset) using Dice score, 95% Hausdorff distance (HD95), and average symmetric surface distance (ASSD).
Main Results:
- The method achieved high Dice scores: 0.818 ± 0.075 for LV and 0.827 ± 0.084 for CP on the BCP dataset.
- On the in-house dataset, Dice scores were even higher: 0.964 ± 0.060 for LV and 0.932 ± 0.059 for CP.
- Quantitative validation using HD95 and ASSD demonstrated the method's accuracy and consistency across both datasets.
Conclusions:
- The proposed deep learning method effectively segments LV and CP in infant brain MRI, overcoming limitations of previous approaches.
- The method ensures anatomical consistency without requiring manual annotation, making it suitable for large-scale studies.
- This technique has the potential to significantly advance research into CP morphology and its role in early neurodevelopment.
Introduction:
Accurate segmentation of the lateral ventricles (LV) and choroid plexus (CP) in infant brain MRI is essential for understanding cerebrospinal fluid dynamics and early neurodevelopment. However, segmentation methods recently introduced for adult populations often underperform on infant data because of rapid anatomical changes, low tissue contrast, and motion artifacts, and they frequently misclassify tissue boundaries.
Method:
To address these challenges, we propose a fully automated deep learning method for joint LV and CP segmentation using T1-weighted MRI (Baby Connectome Project (BCP) dataset total n = 154; in-house retrospective dataset n = 52). Our approach integrates an anatomy-aware loss function that explicitly enforces the topological constraint of CP containment within the LV. The method was validated on two independent datasets to demonstrate clinical adaptability using Dice score, 95% Hausdorff distance (HD95), and the average symmetric surface distance (ASSD).
Results:
The method achieved Dice scores of 0.818 ± 0.075 for LV and 0.827 ± 0.084 for CP in the BCP dataset, with HD95 of 7.487 ± 7.351 mm and 4.925 ± 2.897 mm, and ASSD of 1.175 ± 0.524 mm and 0.818 ± 0.239 mm, respectively. In the in-house dataset, the method achieved Dice scores of 0.964 ± 0.060 for LV and 0.932 ± 0.059 for CP, with HD95 of 0.310 ± 0.542 mm and 4.148 ± 3.726 mm, and ASSD of 0.088 ± 0.151 mm and 0.280 ± 0.239 mm, respectively.
Conclusion:
This method addresses limitations of prior methods by ensuring anatomical consistency without manual annotation. The approach has the potential to support large-scale studies investigating CP morphology and its relevance to early neurodevelopment.
