Related Experiment Video
Updated: Aug 6, 2025
08:40
Quantitation of Protein Expression and Co-localization Using Multiplexed Immuno-histochemical Staining and Multispectral Imaging
Published on: April 8, 2016
12.9K
Development and clinical utility analysis of a prostate zonal segmentation model on T2-weighted imaging: a
Lili Xu1,2, Gumuyang Zhang1, Daming Zhang1
1Department of Radiology, State Key Laboratory of Complex Severe and Rare Disease, Peking Union Medical College Hospital, Peking Union Medical College, Chinese Academy of Medical Sciences, Shuaifuyuan No.1, Wangfujing Street, Dongcheng District, Beijing, 100730, China.
Insights Into Imaging
|March 17, 2023
Summary
A deep learning model accurately segments prostate zones on MRI, outperforming radiologists in some cases. Performance is influenced by prostate size and MRI scanner type.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Accurate segmentation of prostate central gland (CG) and peripheral zone (PZ) is crucial for diagnosis and treatment planning.
- Manual segmentation is time-consuming and subject to inter-observer variability.
Purpose of the Study:
- To develop and validate a deep learning model for automated prostate CG and PZ segmentation on T2-weighted MRI.
- To assess the clinical utility of the model by comparing it with radiologist performance and analyzing influencing factors.
Main Methods:
- A 3D U-Net deep learning model was trained on 223 patients and tested on internal (n=93) and external datasets (n=141, n=59).
- Performance was evaluated using Dice Similarity Coefficients (DSC), 95th Hausdorff Distance (95HD), and Average Boundary Distance (ABD).
- Model performance was compared to a junior radiologist and analyzed for influencing factors like prostate volume and MRI vendor.
Main Results:
- The 3D U-Net model achieved high DSC values for CG (0.869-0.909) and PZ (0.755-0.844) across all datasets.
- The model outperformed a junior radiologist in PZ segmentation (DSC 0.769 vs. 0.706).
- Prostate volume and MRI vendor were identified as significant factors affecting segmentation accuracy.
Conclusions:
- The 3D U-Net model demonstrates robust performance for automated prostate zonal segmentation.
- The model offers a valuable tool for clinical practice, potentially improving efficiency and consistency.
- Segmentation accuracy is influenced by patient-specific factors and imaging parameters, necessitating further research.
