TQGDNet: Coronary artery calcium deposit detection on computed tomography
Related Experiment Video
Updated: May 28, 2025
Identifying Coronary Artery Calcification on Non-gated Computed Tomography Scans
Published on: August 28, 2018
Wei-Chien Wang1, Christopher Yu2, Euijoon Ahn3
1Biomedical Data Analysis and Visualisation Lab, School of Computer Science, The University of Sydney, Australia.
Insights
This study introduces a novel deep learning model for improved coronary artery calcium (CAC) scoring from CT scans. The new model enhances detection of small calcium deposits, leading to more accurate cardiovascular risk assessment.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiology
Background:
- Coronary artery disease (CAD) is a major cause of mortality.
- Coronary artery calcium (CAC) scoring via CT images aids in assessing cardiovascular risk in asymptomatic individuals.
- Current automated CAC scoring methods using convolutional neural networks (CNNs) struggle with integrating features from lower network layers, crucial for detecting small calcium deposits.
Purpose of the Study:
- To develop a new CNN model for enhanced CAC scoring, specifically designed to capture features from small regions and low-contrast areas in CT images.
- To improve the integration of features across multiple CNN layers, particularly focusing on lower layers vital for detecting subtle calcifications.
Main Methods:
- Proposed a novel CNN architecture incorporating a low-contrast detection module (ThrConvs) and two fusion modules: Queen-fusion (Qf) for cross-scale feature fusion and a lower-layer Gather-and-Distribute (GD) module.
- The ThrConvs module uses three convolution blocks for low-contrast object detection.
- The fusion modules connect neurons across adjacent CNN levels to capture comprehensive features of small calcium deposits and their surroundings.
Main Results:
- The proposed model demonstrated superior performance on the public OrCaScore dataset (269 calcium deposits).
- Achieved a 2.3-3.6% improvement in mean Pixel Accuracy (mPA) on both the private Concord and public OrCaScore datasets compared to existing methods.
- Outperformed previous state-of-the-art detection methods in CAC scoring.
Conclusions:
- The novel CNN model effectively captures features from small regions and low-contrast areas, leading to improved CAC scoring accuracy.
- The integration of specialized modules (ThrConvs, Qf, GD) enhances the model's ability to detect subtle coronary artery calcifications.
- This approach offers a significant advancement in automated cardiovascular risk assessment using CT imaging.
Abstract:
Coronary artery disease (CAD) continues to be a leading global cause of cardiovascular related mortality. The scoring of coronary artery calcium (CAC) using computer tomography (CT) images is a diagnostic instrument for evaluating the risk of asymptomatic individuals prone to atherosclerotic cardiovascular disease. State-of-the-art automated CAC scoring methods rely on large annotated datasets to train convolutional neural network (CNN) models. However, these methods do not integrate features across different levels and layers of the CNN, particularly in the lower layers where important information regarding small calcium regions are present. In this study, we propose a new CNN model specifically designed to effectively capture features associated with small regions and their surrounding areas in low-contrast CT images. Our model integrates a specifically designed low-contrast detection module and two fusion modules focusing on the lower layers of the network to connect more deeper and wider neurons (or nodes) across multiple adjacent levels. Our first module, called ThrConvs, includes three convolution blocks tailored to detecting objects in images characterized by low contrast. Following this, two fusion modules are introduced: (i) Queen-fusion (Qf), which introduces a cross-scale feature method to fuse features from multiple adjacent levels and layers and, (ii) lower-layer Gather-and-Distribute (GD) module, which focuses on learning comprehensive features associated with small-sized calcium deposits and their surroundings. We demonstrate superior performance of our model using the public OrCaScore dataset, encompassing 269 calcium deposits, surpassing the capabilities of previous state-of-the-art works. We demonstrate the enhanced performance of our approach, achieving a notable 2.3-3.6 % improvement in mean Pixel Accuracy (mPA) on both the private Concord dataset and the public OrCaScore dataset, surpassing the capabilities of established detection methods.
More Related Videos
06:57Author Spotlight: Advancing Cardiovascular Imaging - Introducing the Spatially Weighted Calcium Score for Early Disease Detection
Published on: September 22, 2023
08:02Author Spotlight: Enhanced Quantification of Cardiovascular Calcification Progression for Longitudinal Micro PET/CT Studies in Small Research Animals
Published on: November 15, 2024
