Abstract:This work aimed to automatically segment and classify the coronary arteries with either normal or anomalous origin from the aorta (AAOCA) using convolutional neural networks (CNNs), seeking to enhance and fasten clinician diagnosis. We implemented three single-view 2D Attention U-Nets with 3D view integration and trained them to automatically segment the aortic root and coronary arteries of 124 computed tomography angiographies (CTAs), with normal coronaries or AAOCA. Furthermore, we automatically classified the segmented geometries as normal or AAOCA using a decision tree model. For CTAs in the test set ( n = 13), we obtained median Dice score coefficients of 0.95 and 0.84 for the aortic root and the coronary arteries, respectively. Moreover, the classification between normal and AAOCA showed excellent performance with accuracy, precision, and recall all equal to 1 in the test set. We developed a deep learning-based method to automatically segment and classify normal coronary and AAOCA. Our results represent a step towards an automatic screening and risk profiling of patients with AAOCA, based on CTA.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the automatic segmentation and classification of the normal origin of the coronary arteries or the abnormal aortic origin of the coronary arteries (AAOCA), using convolutional neural networks (CNNs) to improve the speed and accuracy of clinical diagnosis. Specifically, the researchers developed a deep - learning - based method to automatically segment the aortic root and coronary arteries in CT angiography (CTA) by combining multi - view 2D attention U - Net, and used a decision - tree model to classify these segmented geometric structures as normal or AAOCA. ### Main Objectives: 1. **Automatic Segmentation**: Achieve the automatic segmentation of the aortic root and coronary arteries from CTA images. 2. **Automatic Classification**: Automatically identify whether the origin of the coronary arteries is normal or abnormal based on the segmentation results. 3. **Improve Diagnostic Efficiency**: Reduce the workload of doctors and accelerate the diagnostic process through automated tools. ### Research Background: - **AAOCA**: The coronary artery with an abnormal aortic origin is a rare congenital disease that may lead to sudden cardiac death (SCD) in young people and other serious cardiac events. - **Application of CTA**: Enhanced CT angiography has become the preferred tool for evaluating adult chest pain or myocardial infarction, but manual segmentation and analysis of CTA images are time - consuming and highly subjective. - **Advantages of Deep Learning**: Deep - learning techniques have shown great potential in medical image processing, especially in classification, segmentation, and detection tasks. ### Methods: - **Dataset**: The study used 124 CTA scans, including 50 normal coronary arteries and 74 AAOCA cases. - **Pre - processing**: CTA images were adjusted to a fixed size, and the Hounsfield unit (HU) window was set. - **Network Architecture**: Three single - view 2D attention U - Nets were implemented, corresponding to the axial, sagittal, and coronal planes respectively. - **3D Integration**: By averaging the prediction results of the three planes, the final 3D label map was generated. - **Classification Method**: The segmentation results were classified using 3D connectivity and a decision - tree model. ### Results: - **Segmentation Performance**: In the test set, the median Dice similarity coefficient (DSC) of the aortic root was 0.95, and the median DSC of the coronary arteries was 0.84. - **Classification Performance**: In the test set, the accuracy, precision, and recall of the classification were all 1. ### Discussion: - **Differences in Segmentation Performance**: The segmentation performance of the coronary arteries was slightly lower than that of the aortic root, possibly because the dataset included some rare AAOCA types, such as the posterior aortic origin. - **Future Work**: A larger dataset is required to further validate and improve the model, especially for different types of AAOCA. ### Conclusion: The study successfully developed a deep - learning - based workflow that can automatically segment and classify the aortic root and coronary arteries in CTA images with high accuracy and has good clinical application prospects.

Automatic 3D Segmentation and Identification of Anomalous Aortic Origin of the Coronary Arteries Combining Multi-view 2D Convolutional Neural Networks

Fully automated detection of anomalous aortic origin of the coronary arteries in coronary CT images: a novel artificial intelligence-based screening tool

Automated coronary artery atherosclerosis detection and weakly supervised localization on coronary CT angiography with a deep 3-dimensional convolutional neural network

Segmentation of Aorta 3D CT Images Based on 2D Convolutional Neural Networks

Multi-Resolution 3D Convolutional Neural Networks for Automatic Coronary Centerline Extraction in Cardiac CT Angiography Scans

Multi-Task Deep Convolutional Neural Network for the Segmentation of Type B Aortic Dissection

Automated Coronary Artery Segmentation in Coronary Computed Tomography Angiography (CCTA) Using Deep Learning Neural Networks

An automatic deep learning approach for coronary artery calcium segmentation

Automatic coronary artery segmentation and diagnosis of stenosis by deep learning based on computed tomographic coronary angiography

Fully-automatic segmentation of coronary artery using growing algorithm

A CT-based deep learning system for automatic assessment of aortic root morphology for TAVI planning

Automated Characterization of Stenosis in Invasive Coronary Angiography Images with Convolutional Neural Networks

Deep Learning based CT-scan Coronary Artery Segmentation and Calcium Scoring

CAU-net: A Novel Convolutional Neural Network for Coronary Artery Segmentation in Digital Substraction Angiography.

Automatic Segmentation of Thoracic Aorta Segments in Low-Dose Chest CT

Automatic extraction of coronary arteries using deep learning in invasive coronary angiograms

Automatic extraction and stenosis evaluation of coronary arteries in invasive coronary angiograms

A U-Shaped Network Based on Multi-level Feature and Dual-Attention Coordination Mechanism for Coronary Artery Segmentation of CCTA Images

Context Aware 3D Fully Convolutional Networks for Coronary Artery Segmentation.

Automatic Segmentation of Type A Aortic Dissection on Computed Tomography Images Using Deep Learning Approach

Automated Segmentation of Coronary Arteries Based on Statistical Region Growing and Heuristic Decision Method