Related Experiment Video
Updated: Jan 25, 2026
High-resolution In Vivo Manual Segmentation Protocol for Human Hippocampal Subfields Using 3T Magnetic Resonance Imaging
Published on: November 10, 2015
Dilated Dense U-Net for Infant Hippocampus Subfield Segmentation
Hancan Zhu1,2, Feng Shi3, Li Wang1
1Department of Radiology and Biomedical Research Imaging Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States.
Insights
This study introduces DUnet and ResDUnet, novel deep learning models for segmenting infant brain structures. These methods significantly improve the accuracy of infant hippocampal subfield segmentation from MRI scans.
Area of Science:
- Medical Imaging
- Neuroscience
- Artificial Intelligence
Background:
- Accurate segmentation of infant hippocampal subfields is crucial for understanding memory-related neurological disorders.
- Existing segmentation methods, developed for adults, perform poorly on infant MR images due to developmental variations and low tissue contrast.
Purpose of the Study:
- To develop and evaluate novel deep learning models for accurate automatic segmentation of infant hippocampal subfields.
- To address the limitations of current methods in handling the unique challenges of infant brain imaging.
Main Methods:
- Proposed a novel fully convolutional network (FCN) named DUnet, integrating a dilated dense network within the U-net architecture.
- Further enhanced DUnet by incorporating residual connections to create Residual DUnet (ResDUnet).
- These architectures leverage multi-scale features while maintaining high spatial resolution for improved feature fusion.
Main Results:
- DUnet and ResDUnet demonstrated improved average Dice coefficients of 2.1% and 2.5%, respectively, compared to the standard 3D U-net for infant hippocampal subfield segmentation.
- The proposed methods outperformed other state-of-the-art segmentation techniques.
Conclusions:
- DUnet and ResDUnet represent significant advancements in the automatic segmentation of infant hippocampal subfields.
- These models offer improved accuracy and robustness for analyzing infant brain development and related neurological conditions.
Abstract:
Accurate and automatic segmentation of infant hippocampal subfields from magnetic resonance (MR) images is an important step for studying memory related infant neurological diseases. However, existing hippocampal subfield segmentation methods were generally designed based on adult subjects, and would compromise performance when applied to infant subjects due to insufficient tissue contrast and fast changing structural patterns of early hippocampal development. In this paper, we propose a new fully convolutional network (FCN) for infant hippocampal subfield segmentation by embedding the dilated dense network in the U-net, namely DUnet. The embedded dilated dense network can generate multi-scale features while keeping high spatial resolution, which is useful in fusing the low-level features in the contracting path with the high-level features in the expanding path. To further improve the performance, we group every pair of convolutional layers with one residual connection in the DUnet, and obtain the Residual DUnet (ResDUnet). Experimental results show that our proposed DUnet and ResDUnet improve the average Dice coefficient by 2.1 and 2.5% for infant hippocampal subfield segmentation, respectively, when compared with the classic 3D U-net. The results also demonstrate that our methods outperform other state-of-the-art methods.
Related Concept Videos
01:13Dense Connective Tissue
Dense Regular Connective Tissue
In dense regular connective tissue, fibers are arranged parallel to each other, enhancing its tensile strength and resistance to stretching in the direction of the fiber orientations. Ligaments and tendons are made of dense regular...
01:30Cardiomyopathy II: Dilated Cardiomyopathy
01:19Role of Hippocampus in Memory
01:29Drug Dosing: Infants and Children
03:34Theory of Romantic Attachment in Adulthood
01:07What is Weather?