Related Experiment Video
Updated: May 2, 2026
11:27
Magnetically-Assisted Remote Controlled Microcatheter Tip Deflection under Magnetic Resonance Imaging
Published on: April 4, 2013
12.4K
Rapid multi-catheter segmentation for magnetic resonance image-guided catheter-based interventions
Amanda M Aleong1, Alejandro Berlin2,3, Jette Borg2,3
1Institute of Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada.
Medical Physics
|May 7, 2024
Summary
This study developed an AI algorithm for segmenting multiple catheters in MRI scans during interventions. The AI achieved high accuracy, comparable to human experts, with minimal impact on treatment planning.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Interventional Radiology
Background:
- Magnetic resonance imaging (MRI) is crucial for soft tissue lesion delineation.
- Accurate catheter placement is vital for catheter-based interventions.
- Manual segmentation of catheters in MRI is time-consuming and challenging.
Purpose of the Study:
- To develop and evaluate a machine learning algorithm for automated catheter detection in MRI.
- To improve the efficiency of MR-guided procedures through automated segmentation.
Main Methods:
- A 3D U-Net model was trained for catheter segmentation in MR images from prostate brachytherapy.
- Multiple AI models were trained to assess segmentation confidence.
- Performance was evaluated against manual segmentations and dosimetric impact on treatment plans.
Main Results:
- The AI successfully identified 95% of catheters with rapid processing (0.89s per volume).
- Multi-model approach identified cases needing user input, ensuring reliability.
- AI segmentation accuracy was comparable to independent observers, with similar plan dosimetry.
Conclusions:
- AI-driven catheter segmentation is highly accurate and efficient for MR-guided interventions.
- Multi-AI model approach enhances confidence and identifies segmentation requiring manual review.
- Real-time AI segmentation can aid in assessing catheter deflection and expedite prostate brachytherapy planning.
