Multi-view Three-Dimensional Fusion Echocardiography Using a Novel Respiratory Tracking Technique: First Results in Humans

• Author / Creator

  Lamb, Tyler

• Introduction: 3-dimensional echocardiography (3DE) is currently the echocardiographic method of choice for quantification of left ventricular systolic function according to the American Society of Echocardiography. However, while developments in ultrasound transducer technology and post-processing techniques have undoubtedly bettered 3DE, they have failed to address inherent weaknesses of 3DE stemming from the nature of ultrasound physics. These include a limited field-of-view (FOV), reduced spatial and temporal resolution and reduced endocardial border definition (EBD) as compared to 2-dimensional echocardiography (2DE). In the context of echocardiography in general, poor EBD is largely explained by weakly reflected signals from important interfaces like the left ventricular (LV) endocardial border that often result from non-perpendicular angles of insonation. A technique called ‘multi-view 3D fusion echocardiography’ (M3DFE) provides a solution to this dilemma by fusing 3DE datasets from various complementary acoustic windows.
  
  The thesis begins with a review of the literature around M3DFE and related topics. M3DFE has been previously studied in pre-clinical settings and has demonstrated favorable results by a number of investigators. Considering the existing literature, it is reasonable to infer that the most promising approach to M3DFE involves an optical tracking technique. Here, spatial alignment of datasets is accomplished through the optical tracking of both chest and transducer markers. However, two main challenges have not been adequately addressed: i) no clinically feasible alignment protocol exists, and ii) there is no consensus on the optimal way to process overlapping portions of M3DFE datasets.
  
  This thesis examines two hypotheses which, if proven true, have the potential to improve the feasibility and effectiveness of M3DFE and bring it closer to readiness for clinical testing. First, we hypothesize that a novel respiratory tracking technique based on quantitative optical tracking of chest markers can optimize alignment of datasets such that >90% will be suitable for diagnostic assessment as validated by an absence of perceptible misalignment. Second, we hypothesize that a fusion technique based on wavelet decomposition is superior to a more basic technique based on voxel averaging.
  Methods: 3D Real-time M3DFE datasets were acquired from eleven volunteers during a breath-hold maneuver with three imaging protocols: i) using an unmoving transducer capturing a standard apical view, ii) using slight movements of the transducer to include non-standard apical views, and iii) with the transducer positioned at both apical and parasternal windows. Infrared cameras were used to track the 3D position and orientation of the transducer and chest markers. Chest marker tracking data was used to perform a novel quantitative screening procedure aimed at predicting adequate alignment of datasets. Multi-planar reconstruction of both M3DFE and standard apical 3DE datasets was performed to generate four- and two- chamber 2-dimensional planes which were then subjected to both fusion by voxel averaging and wavelet decomposition and compared. Subjective assessments included i) successful alignment of datasets and ii) EBD. Objective assessments included i) contrast, ii) contrast-to-noise ratio, iii) signal-to-noise ratio and iv) % increase in FOV.
  Results: The quantitative screening procedure was effective and yielded a 97% rate of accurately predicting subjective alignment. Both fusion by voxel averaging and wavelet decomposition improved subjective and objective measures of image quality and FOV. Wavelet decomposition was generally superior to voxel averaging with respect to contrast and EBD in all three imaging protocols, and with respect to SNR in the apical-parasternal protocol.
  Conclusion: Our novel screening procedure based on quantitative optical tracking of chest markers is effective at optimizing alignment of M3DFE datasets. Results were generally supportive of the hypothesis that fusion by wavelet decomposition is a superior method to fusion by voxel averaging.

• Subjects / Keywords

  • 3D Echocardiography

• Graduation date

  Fall 2018

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/R3VQ2SS4C

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.