Related Experiment Video
Updated: Aug 4, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
DBCU-Net: deep learning approach for segmentation of coronary angiography images
Yuqiang Shen1, Zhe Chen2, Jijun Tong2
1The Fourth Affiliated Hospital Zhejiang University School of Medicine, Jinhua, China.
Insights
We developed DBCU-Net, a novel deep learning model, for automatic coronary artery segmentation in coronary angiography (CAG) images. This method improves the accuracy of blood vessel segmentation, aiding in coronary artery disease diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiovascular Disease Research
Background:
- Coronary angiography (CAG) is the gold standard for diagnosing coronary artery disease (CAD).
- Current CAG imaging methods suffer from low resolution, poor contrast, artifacts, and noise, hindering accurate blood vessel segmentation.
- Precise segmentation is crucial for quantitative analysis and diagnosis of CAD.
Purpose of the Study:
- To propose a novel deep learning model, DBCU-Net, for automated and accurate segmentation of coronary arteries in CAG images.
- To enhance feature extraction in CAG images by integrating dense connectivity and bi-directional ConvLSTM (BConvLSTM).
- To overcome the limitations of traditional convolutional approaches in U-Net architectures for medical image segmentation.
Main Methods:
- Development of DBCU-Net, an extended U-Net architecture incorporating DenseNet and bi-directional ConvLSTM (BConvLSTM) layers.
- Utilizing dense connectivity and BConvLSTM for improved feature highlighting and extraction within the network.
- Experimental validation conducted on a private dataset of coronary angiography images.
Main Results:
- The DBCU-Net achieved high performance metrics for coronary artery segmentation.
- Average Accuracy: 0.985
- Average Precision: 0.913
- Average Recall: 0.847
- Average F1-score: 0.879
Conclusions:
- DBCU-Net demonstrates significant potential for accurate automatic segmentation of coronary arteries in CAG images.
- The integration of dense connectivity and BConvLSTM effectively enhances feature representation for improved segmentation.
- This approach offers a promising tool for aiding in the diagnosis and management of coronary artery disease.
Abstract:
Coronary angiography (CAG) is the "gold standard" for diagnosing coronary artery disease (CAD). However, due to the limitation of current imaging methods, the CAG image has low resolution and poor contrast with a lot of artifacts and noise, which makes it difficult for blood vessels segmentation. In this paper, we propose a DBCU-Net for automatic segmentation of CAG images, which is an extension of U-Net, DenseNet with bi-directional ConvLSTM(BConvLSTM). The main contribution of our network is that instead of convolution in the feature extraction of U-Net, we incorporate dense connectivity and the bi-directional ConvLSTM to highlight salient features. We conduct our experiment on our private dataset, and achieve average Accuracy, Precision, Recall and F1-score for coronary artery segmentation of 0.985, 0.913, 0.847 and 0.879 respectively.
