A model based approach to improve the performance of the geometric filtering speckle reduction algorithm

This paper presents the methods used to adapt the geometric filtering method for speckle reduction to ultrasound imaging. The geometric filtering method is an iterative algorithm for speckle reduction which was first applied to radar images obtained with well controlled axial and lateral resolution. The appearance of speckle in ultrasound images is directly related to the size of the point-spread-function which is known to vary through-out a single frame. In order to optimally apply the speckle reduction algorithm in ultrasound, the effects of transducer geometry, center frequency shifts, and beamforming geometry were modeled and used to resample either the raw or video data before speckle processing. As a result of this approach, less data needs to be processed and the number of iterations are reduced. Using commercially available signal processing hardware, speckle reduced ultrasound frames can be processed and displayed at rates approaching 2 per second. This improvement in throughput facilitates the clinical evaluation of the geometric filtering method for improving lesion detection and overall image interpretability