Speckle Reduction and Edge Detection in Ultrasound Imagery
|dc.description.abstract||This thesis investigates the related topics of speckle reduction and edge detection as applied to speckled ultrasonography. A strategy for the evaluation of speckle filters is presented. This includes their application to a large test set of both clinical and simulated images. Functional performance is measured by application of objective image quality metrics, and the relationship between these objective metrics and the subjective opinion of clinical experts is investigated. Finally a detailed analysis of the complexity of the filtering methods is performed to ascertain the computational requirements of the different filters. Edge detection is investigated using Gabor filters to perform orientation-specific multiscale decompositions. The effect of speckle is reduced in the Gabor transforms by MAP estimation. This involves modelling the statistics of the speckle Gabor transform as a combined Gaussian and symmetric alpha stable distribution. A technique is proposed for the estimation of the parameters of this model, and its accuracy is demonstrated. Edge detection is achieved by finding the zero crossings of the even-symmetric Gabor coefficients. To suppress the zero crossings which do not correspond to images edges, two observed properties of the Gabor zero crossings are examined. These measurements are combined to form the final edge detector, which is shown to exceed the performance of other edge detectors for speckled imagery. A close relationship exists between edge detection and speckle removal, especially for the diffusion class of filters. The proposed zero crossing edge detector is accordingly incorporated into a tensor-valued diffusion scheme. The performance of this speckle filter is evaluated using the framework proposed in this thesis, and is found to exhibit excellent functional performance.||en_US|
|dc.title||Speckle Reduction and Edge Detection in Ultrasound Imagery||en_US|
|dc.local.note||This thesis investigates some of the challenges in analysis of echocardiographic (cardiac ultrasound) images. In particular, the presence of a particular form of image noise called speckle reduces image quality and usefulness. A large number of methods of reducing speckle are evaluated using a comprehensive comparison strategy. A new method is proposed based on statistical modeling and edge detection, and is shown to perform better than existing approaches.||en_US|
Files in this item
This item is available under the Attribution-NonCommercial-NoDerivs 3.0 Ireland. No item may be reproduced for commercial purposes. Please refer to the publisher's URL where this is made available, or to notes contained in the item itself. Other terms may apply.
The following license files are associated with this item: