Related Experiment Video
Updated: Jan 15, 2026
06:46
Quantitative Micro-CT Analysis of Aortopathy in a Mouse Model of β-aminopropionitrile-induced Aortic Aneurysm and Dissection
Published on: July 16, 2018
10.0K
Automatic Segmentation of Abdominal Aortic Aneurysm From Computed Tomography Angiography Using a Patch-Based Dilated
Merjulah Roby1, Juan C Restrepo1, Haehwan Park2
1Department of Mechanical Engineering, The University of Texas at San Antonio, San Antonio, TX 78249, USA.
IEEE Access : Practical Innovations, Open Solutions
|October 6, 2025
Summary
Automated segmentation of Abdominal Aortic Aneurysm (AAA) using deep learning and NURBS significantly improves accuracy and speed for clinical applications. This advancement aids in faster, more precise treatment planning for AAA patients.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiovascular Surgery
Background:
- Abdominal Aortic Aneurysm (AAA) presents a significant public health challenge with rising mortality rates.
- Computed Tomography Angiography (CTA) is crucial for AAA monitoring and surgical planning.
- Manual segmentation of CTA images is time-consuming and labor-intensive, necessitating automated solutions.
Purpose of the Study:
- To develop and evaluate an automated deep learning framework for segmenting AAA in CTA images.
- To enhance segmentation accuracy and efficiency for clinical decision-making in AAA management.
Main Methods:
- A patch-based dilated modified U-Net deep learning model was adapted for automated AAA segmentation.
- Non-Uniform Rational B-Splines (NURBS) were integrated to refine segmentation accuracy.
- The model's processing speed was evaluated during the prediction phase.
Main Results:
- The deep learning framework achieved accurate delineation of AAA regions in CTA scans.
- The model demonstrated exceptional processing speed, requiring only 17 ± 0.02 milliseconds per frame.
- NURBS integration significantly improved segmentation accuracy for intricate anatomical contours.
Conclusions:
- The proposed automated segmentation model offers a promising clinical tool for efficient and accurate AAA management.
- Integration of deep learning with NURBS provides a robust solution for the clinical need in medical image segmentation.
- The fast processing time and high accuracy support its application in real-time clinical workflows for AAA patients.
