Related Experiment Video
Updated: Jul 20, 2025
09:41
A Pipeline for 3D Multimodality Image Integration and Computer-assisted Planning in Epilepsy Surgery
Published on: May 20, 2016
12.4K
Suitability of DNN-based vessel segmentation for SIRT planning
Farina Kock1, Felix Thielke2, Nasreddin Abolmaali3
1Fraunhofer Institute for Digital Medicine MEVIS, Max-von-Laue-Str. 2, Bremen, 28359, Germany. farina.kock@mevis.fraunhofer.de.
Summary
Deep learning segmentation of hepatic arteries shows promise for selective internal radiation therapy planning. This AI approach can potentially speed up pre-interventional planning for liver cancer treatment.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Hepatic artery (HA) segmentation is crucial for planning selective internal radiation therapy (SIRT), a liver cancer treatment.
- SIRT involves catheter-based delivery of radioactive beads to liver tumors.
- Accurate HA segmentation is vital for effective and safe radioembolization.
Purpose of the Study:
- To evaluate a deep neural network (DNN) for segmenting hepatic arteries for SIRT planning.
- To compare the DNN segmentation performance against a traditional machine learning algorithm.
- To assess the clinical suitability of DNN-based HA segmentation for pre-interventional SIRT planning.
Main Methods:
- Applied a DNN-based HA segmentation on 36 contrast-enhanced CT scans.
- Rated segmentation and image quality of the DNN approach.
- Compared DNN results with a traditional machine learning algorithm.
- Conducted expert ratings on the usability of segmentations for SIRT planning.
Main Results:
- The DNN approach outperformed the traditional machine learning algorithm.
- DL segmentation was suitable for SIRT planning in a significant portion of cases.
- Manual reference segmentations were sufficient in a comparable percentage of cases.
- Some cases showed discrepancies, with DL segmentations being usable when manual ones were not, and vice versa.
Conclusions:
- Hepatic artery segmentation is challenging and time-consuming.
- Deep learning methods offer potential to support and accelerate SIRT pre-interventional planning.
- AI-driven segmentation can improve efficiency and accuracy in liver cancer treatment planning.
