Scholarlink

This article presents a fully automated algorithm for segmenting the Aortic Annulus (AA) in non-contrast Cardiac Magnetic Resonance (CMR) imaging. This advancement is especially significant as it removes the need for contrast-enhanced CT coronary angiography (CCTA), the current gold standard for this task. This is crucial for patients with advanced chronic kidney disease or contrast allergies, where the use of contrast agents is contraindicated. Selecting the appropriate valve graft during the planning of Transcatheter Aortic Valve Implantation (TAVI) relies on the precise measurement of the Aortic Annulus (AA). While numerous software tools are available for AA sizing using CCTA, there is none for CMR imaging. To address this gap, our approach utilizes two segmentation networks. The first network segments the aorta, a well-defined anatomical structure. Then, we apply an unwrapping technique to produce 2D maps of the aorta. These maps serve as input for the second segmentation network, identifying the hinge points necessary to delineate the AA structure and extract key TAVI parameters. Comprehensive evaluations of our algorithm show a Root Mean Square Error (RMSE) of 2.671 mm in the axial direction and a Mean Absolute Error (MAE) of 1.940 mm in AA diameter. Statistical analyses reveal a correlation coefficient of 0.782 for AA diameter between the manual method and our method. A blind test further supports its reliability. Our algorithm proposes automating CMR-based AA segmentation to advance patient-friendly and low-risk imaging modalities.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Beyond the Standards: Fully-Automated Aortic Annulus Segmentation on Contrast-Free Magnetic Resonance Imaging Using a Computational Aorta Unwrapping Method

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Enrique Almar-Munoz,
  • Mathias Pamminger,
  • Christian Kremser,
  • Markus Haltmeier,
  • Agnes Mayr

摘要

This article presents a fully automated algorithm for segmenting the Aortic Annulus (AA) in non-contrast Cardiac Magnetic Resonance (CMR) imaging. This advancement is especially significant as it removes the need for contrast-enhanced CT coronary angiography (CCTA), the current gold standard for this task. This is crucial for patients with advanced chronic kidney disease or contrast allergies, where the use of contrast agents is contraindicated. Selecting the appropriate valve graft during the planning of Transcatheter Aortic Valve Implantation (TAVI) relies on the precise measurement of the Aortic Annulus (AA). While numerous software tools are available for AA sizing using CCTA, there is none for CMR imaging. To address this gap, our approach utilizes two segmentation networks. The first network segments the aorta, a well-defined anatomical structure. Then, we apply an unwrapping technique to produce 2D maps of the aorta. These maps serve as input for the second segmentation network, identifying the hinge points necessary to delineate the AA structure and extract key TAVI parameters. Comprehensive evaluations of our algorithm show a Root Mean Square Error (RMSE) of 2.671 mm in the axial direction and a Mean Absolute Error (MAE) of 1.940 mm in AA diameter. Statistical analyses reveal a correlation coefficient of 0.782 for AA diameter between the manual method and our method. A blind test further supports its reliability. Our algorithm proposes automating CMR-based AA segmentation to advance patient-friendly and low-risk imaging modalities.