AbstractEchocardiography (echo) is a widely available method to obtain images of the heart, however, echo can suffer due to the presence of artefacts, high noise and a restricted field-of-view. One method to overcome these limitations is to use multiple images, using the "best" parts from each image to produce a higher quality "compounded" image. This thesis describes a new method to allow multiple 3D echo images to be compounded into a single better quality volume. I have proposed a definition for an "ideal" compounded image and have used this to guide the design of my compounding method, in particular designing a method to reduce the effect of image artefacts and to make use of larger numbers of images. My compounding method has been validated using phantom, volunteer and clinical images. The overall motivations for improving echo image quality are twofold: Firstly to provide clinicians with higher quality images which I hope will improve the accuracy of clinical decision making. Secondly to provide higher quality images for subsequent post-processing algorithms. A number of methods have been proposed to compound sets of ultra-sound images, all of which have reported improvements in image quality. However, previous 3D compounding methods have typically been applied to a relatively small number of images (most of them only use two images, and only one uses six images). I have investigated the effect of compounding with larger numbers of images. Results showed continued improvement in image quality up to ten images (the maximum number we deemed feasible to acquire in a clinical setting and it is approximately double of images used previously). Artefacts occur regularly within echo images, particularly shadowing artefacts (due to the highly reflecting interfaces caused by the ribs and lungs when imaging the heart). However, previous 3D compounding methods haven’t directly claimed and demonstrated the effect of artefacts.
Therefore, I have proposed a 3D compounding algorithm which specifically aims to reduce the effect of echo artefacts (shadowing) as well as improving the signal-to-noise ratio, contrast, and extending the field-of-view. My method to reduce the effect of artefacts is to weight image information from different views based on a local feature coherence/consistency. I hypothesize that the presence of an artefact in an image varies greatly depending on view direction, therefore much lower consistency values will be calculated for artefact regions enabling them to be detected, and their influence on the compounded image to be greatly reduced. The accuracy of the image registration is important and errors will likely affect the final compounded images quality. In addition to registration ac-curacy my system needs to work robustly and have a large enough capture range to enable automatic registration from a suitable starting position.
|Date of Award||2012|
|Supervisor||Graeme Penney (Supervisor) & Tobias Schaeffter (Supervisor)|