Related Experiment Video
Updated: Jun 13, 2025
04:25
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
2.3K
Morphology and Texture-Guided Deep Neural Network for Intracranial Aneurysm Segmentation in 3D TOF-MRA
Maysam Orouskhani1, Negar Firoozeh1, Huayu Wang1,2
1Department of Radiology, University of Washington, Seattle, WA, USA.
Neuroinformatics
|September 11, 2024
Summary
This study introduces a new method to improve the segmentation of intracranial aneurysms by reweighting morphology and texture in deep learning models. This approach enhances accuracy for better diagnosis and treatment planning.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Neurosurgery
Background:
- Intracranial aneurysms pose significant diagnostic and treatment challenges.
- Accurate segmentation is crucial for clinical decision-making.
- Existing methods struggle with instance imbalance and morphological variability.
Purpose of the Study:
- To develop a novel approach for intracranial aneurysm segmentation.
- To address challenges of instance imbalance and morphological variability.
- To improve the accuracy and reliability of aneurysm segmentation.
Main Methods:
- Introduced a morphology and texture loss reweighting approach for deep neural networks.
- Incorporated tailored weights in the loss function to account for aneurysm size, shape, and texture.
- Validated the method using ADAM and RENJI Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) datasets.
Main Results:
- Demonstrated significant improvement in aneurysm segmentation accuracy.
- The proposed method effectively handles imbalanced features and morphological variations.
- Achieved promising outcomes for enhanced diagnostic insights.
Conclusions:
- The morphology and texture loss reweighting approach offers a nuanced solution for intracranial aneurysm segmentation.
- Tailored weights and dynamic adaptability enhance segmentation precision.
- The method shows potential for accurate diagnostics and informed treatment strategies in neuroimaging.
