Comparison of techniques for speckle noise reduction in breast ultrasound images

Abstract

Ultrasound scanning is considered the most important complementar method of breast cancer diagnosis. However, the acquisition format of this diagnostic method uses to introduce a remarkable amount of noise in the image, being the speckle noise of primary importance since it gives the image a granular aspect hindering the detection mainly of small objects in the image. With this feature in view, this work attempts to make a comparison among some of techniques proposed with aims of reducing this type of noise, such as: Frost Filter, Lee Filter, Wiener Filter, and Discreet Wavelet Transform. The results are analyzed by a series of evaluation functions which can be used as figures of merit in the image quality evaluation - Root Mean Square Error, Signal to Noise Ratio rms, and Peak Signal to Noise Ratio. The results have indicated that the Wavelet Transform seems to be more efficient in achieving the main purpose of speckle reduction in breast ultrasound images. © 2009 Springer-Verlag.