-
A Deep learning Approach to Generate Contrast-Enhanced Computerised Tomography Angiography without the Use of Intravenous Contrast Agents
Contrast-enhanced computed tomography angiograms (CTAs) are widely used ...
read it
-
D-Net: Siamese based Network with Mutual Attention for Volume Alignment
Alignment of contrast and non-contrast-enhanced imaging is essential for...
read it
-
Aorta Segmentation for Stent Simulation
Simulation of arterial stenting procedures prior to intervention allows ...
read it
-
Multi-Contrast Computed Tomography Healthy Kidney Atlas
The construction of three-dimensional multi-modal tissue maps provides a...
read it
-
Radiologist-level stroke classification on non-contrast CT scans with Deep U-Net
Segmentation of ischemic stroke and intracranial hemorrhage on computed ...
read it
-
Improving Myocardium Segmentation in Cardiac CT Angiography using Spectral Information
Accurate segmentation of the left ventricle myocardium in cardiac CT ang...
read it
-
Graph Attention Network based Pruning for Reconstructing 3D Liver Vessel Morphology from Contrasted CT Images
With the injection of contrast material into blood vessels, multi-phase ...
read it
A Deep Learning Approach to Automate High-Resolution Blood Vessel Reconstruction on Computerized Tomography Images With or Without the Use of Contrast Agent
Existing methods to reconstruct vascular structures from a computed tomography (CT) angiogram rely on injection of intravenous contrast to enhance the radio-density within the vessel lumen. However, pathological changes can be present in the blood lumen, vessel wall or a combination of both that prevent accurate reconstruction. In the example of aortic aneurysmal disease, a blood clot or thrombus adherent to the aortic wall within the expanding aneurysmal sac is present in 70-80 extraction of vital clinically relevant information by current methods. In this study, we implemented a modified U-Net architecture with attention-gating to establish a high-throughput and automated segmentation pipeline of pathological blood vessels in CT images acquired with or without the use of a contrast agent. Twenty-six patients with paired non-contrast and contrast-enhanced CT images within the ongoing Oxford Abdominal Aortic Aneurysm (OxAAA) study were randomly selected, manually annotated and used for model training and evaluation (13/13). Data augmentation methods were implemented to diversify the training data set in a ratio of 10:1. The performance of our Attention-based U-Net in extracting both the inner lumen and the outer wall of the aortic aneurysm from CT angiograms (CTA) was compared against a generic 3-D U-Net and displayed superior results. Subsequent implementation of this network architecture within the aortic segmentation pipeline from both contrast-enhanced CTA and non-contrast CT images has allowed for accurate and efficient extraction of the entire aortic volume. This extracted volume can be used to standardize current methods of aneurysmal disease management and sets the foundation for subsequent complex geometric and morphological analysis. Furthermore, the proposed pipeline can be extended to other vascular pathologies.
READ FULL TEXT
Comments
There are no comments yet.