Biased Temporal Filtering in Ultrasound Imaging
Evans, A.N. and Nixon, M.S. (1994) Biased Temporal Filtering in Ultrasound Imaging.
Full text not available from this repository.
Ultrasonic images are affected by multiplicative speckle noise that reduces the quality of images in which it is present by reducing the contrast, lowering the signal to noise ratio (SNR) and obscuring important diagnostic detail. The intensity distribution for regions corrupted by speckle follows a Rayleigh distribution. Approaches to the reduction of speckle can be divided into three main classes: image filtering, phase-based methods and compounding. In compounding, we seek to exploit the non-stationarity of speckle by combining a series of input images to produce a single, improved output image. Images may be averaged over time, space or frequency. In spatial and frequency compounding simple image addition has not proved successful, resulting in resolution loss. Improved results can be achieved via weighted averaging or by frequency diversity processing, though interest in spatial compounding has waned because it is only applicable to a few sites of clinical interest. In comparison time compounding has been largely ignored as a research area. Direct averaging is used by many ultrasound scanners but results in blurring where features of interest are moving over time. However temporal methods have several intuitive advantages; the frame rate of ultrasound scanners is sufficient to produce a series of images with independent speckle patterns without the need for multi element transducers or multiple frequencies.
|Item Type:||Monograph (Technical Report)|
|Additional Information:||1994 Research Journal Address: Department of Electronics and Computer Science|
|Divisions:||Faculty of Physical Sciences and Engineering > Electronics and Computer Science > Comms, Signal Processing & Control
|Date Deposited:||05 Aug 1999|
|Last Modified:||27 Mar 2014 19:50|
|Further Information:||Google Scholar|
|RDF:||RDF+N-Triples, RDF+N3, RDF+XML, Browse.|
Actions (login required)