Related Experiment Video
Updated: Nov 2, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.1K
Fully automated segmentation of brain tumor from multiparametric MRI using 3D context deep supervised U-Net
Mingquan Lin1, Shadab Momin1, Yang Lei1
1Department of Radiation Oncology and Winship Cancer Institute, Emory University, Atlanta, GA, USA.
Medical Physics
|June 8, 2021
Summary
This study introduces a deep learning model for automatic brain tumor segmentation in MR images, significantly improving accuracy and reducing manual effort. The developed method shows potential for clinical translation in radiotherapy workflows.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuro-oncology
Background:
- Brain tumor diagnosis relies on multimodality imaging for accurate subregion delineation.
- Current manual slice-by-slice segmentation is time-consuming and prone to variability.
- Automating segmentation can enhance diagnostic accuracy and efficiency.
Purpose of the Study:
- To develop an automatic deep learning-based segmentation method for brain tumors in MR images.
- To improve the accuracy and efficiency of brain tumor subregion delineation.
- To reduce intra- and inter-rater variabilities in tumor segmentation.
Main Methods:
- A context deep-supervised U-Net architecture was developed for dense segmentation.
- A context block aggregated multiscale information to enlarge the effective receptive field.
- The model was validated on the BraTS 2020 dataset using fivefold cross-validation.
Main Results:
- The proposed method achieved high segmentation accuracy with Dice Similarity Coefficients (DSC) of 0.923 (WT), 0.893 (TC), and 0.846 (ET).
- The method demonstrated significantly better segmentation accuracy compared to two state-of-the-art CNN networks (p < 0.05).
- High positive correlation was observed between automated and manual tumor volume contours.
Conclusions:
- The developed deep learning technique shows significant potential for clinical application in brain tumor segmentation.
- The method can streamline radiotherapy workflows by providing accurate and efficient tumor delineation.
- Further translation into clinical practice is supported by robust quantitative and qualitative results.
