Ultrasound image deconvolution using adaptive inverse filtering

Clinical ultrasound image quality typically suffers from reduced resolution due to the effects of limited effective aperture size (convolutional blurring), out-of-focus blurring and noise. Speckle noise in ultrasound images can be especially troublesome and significantly deteriorates image quality. This paper discusses the use of a novel approach developed for optical microscope images and applied to ultrasound. In earlier research, 3D microscope images were successfully deconvolved using an adaptive, least-mean-square solution to a statistical Wiener filter. The filter can be applied to two and three-dimensional images as a finite impulse response (FIR) filter solved adaptively using an inverse model of the point-spread-function (PSF). For ultrasound, the filter can be solved using the response to a phantom where the desired result is known a priori. The filter is solved adaptively to minimize the mean-square-error. The resulting filter can then be applied to any image acquired with the same transducer array and instrument parameters. The application of this type of filter to ultrasound is in the preliminary stages with some partial success obtained so far. Preliminary results are shown on phantom and clinical ultrasound data (courtesy of Diasonics, Inc.) processed by this method. The results are promising for improving resolution and minimizing noise