Three-dimensional Ultrasound Fusion for Transesophageal Echocardiography

Publication Type:
Thesis
Issue Date:
2023
Full metadata record
In the past decade, three-dimensional transesophageal echocardiography (3D TEE) has been increasingly used for assessing cardiac anatomy and function. However, 3D TEE suffers from a limited field of view (FoV) of the ultrasound transducer, which makes it difficult to examine complete regions of interest of the heart in a single volume. In this thesis, five innovative multi-view registration and fusion algorithms are proposed to enlarge the FoV of 3D TEE images, with potential applications to assist clinicians with diagnosis, therapy, and surgery for patients with various cardiac diseases. The research starts with achieving efficient pairwise registration and fusion, followed by novel simultaneous registration and fusion methods for improving accuracy and reducing computational complexity. Both the rigid and deformable scenarios are considered in pairwise and simultaneous registration and fusion. This thesis introduces the first intensity-based bundle adjustment framework named direct bundle adjustment (DBA), which can complete the 3D image registration and fusion simultaneously. Furthermore, the DBA framework is extended to solve dynamic simultaneous registration and fusion problems. Rigorous experiments conducted on in-vivo 3D TEE images have demonstrated the effectiveness of the proposed methods for extending the FoV of 3D TEE images and the important clinical value of these methods.
Please use this identifier to cite or link to this item: