Related Experiment Video
Updated: Apr 16, 2026
04:25
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
4.2K
Deep Learning Informed by Contrast-enhanced MRI Allows for Precise Segmentation of the Choroid Plexus in Non-contrast
Araz Incesoy1, Maximilian F Russe2, Jonas A Hosp3
1Department of Neuroradiology, Medical Center - University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany (A.I., H.U., A.R., T.D.).
Academic Radiology
|April 15, 2026
Summary
A new deep learning algorithm accurately segments the choroid plexus (ChP) in brain MRI scans, improving insights into neurological conditions. This automated segmentation method offers a reliable tool for analyzing the choroid plexus in both contrast-enhanced and non-contrast imaging.
Area of Science:
- Neuroimaging
- Artificial Intelligence
- Cerebrospinal Fluid Dynamics
Background:
- The choroid plexus (ChP) plays a crucial role in cerebrospinal fluid (CSF) production and neuroimmunology.
- Accurate delineation of the ChP in neuroimaging is essential but challenging.
- Existing segmentation methods may lack precision for clinical applications.
Purpose of the Study:
- To develop and validate a deep learning-based automated segmentation algorithm for the choroid plexus (ChP).
- To assess the algorithm's performance on both contrast-enhanced (CE) and non-contrast (NC) brain MRI datasets.
- To compare the novel algorithm against an established segmentation technique.
Main Methods:
- A deep neural patchwork (DNP) algorithm was trained using 149 patient datasets with pre- and post-contrast T1w and T2w MRI images.
- Manual segmentation of the lateral ventricle ChP on CE T1w images served as ground truth.
- The DNP was applied to 50 independent datasets and compared with FreeSurfer segmentation on NC T1w images.
Main Results:
- The DNP algorithm achieved a mean Dice coefficient of 0.81 ± 0.05 for ChP segmentation on CE T1w images.
- Satisfactory segmentation quality was obtained on NC T1w (0.71 ± 0.05) and T2w (0.70 ± 0.05) images.
- The DNP significantly outperformed FreeSurfer (0.38 ± 0.05) across all tested contrasts (p < .001).
Conclusions:
- The developed deep learning algorithm enables reliable automated segmentation of the choroid plexus (ChP) in both CE and NC brain MRI.
- This advanced segmentation technique may provide deeper insights into the long-term progression of physiological and pathological processes involving the ChP.
- The DNP algorithm offers a robust alternative to existing methods for ChP analysis in neuroimaging research and clinical practice.
