Speckle reduction in medical ultrasound images

A multiresolution approach using Markov Random Field models

A. Pizurica, W. Philips, I. Lemahieu, and M. Acheroy, "The Application of Markov Random Field Models to Wavelet-Based Image Denoising," Imaging and Vision Systems: Theory, Assessment and Applications, editors J. Blanc-Talon and D. Popescu, NOVA Science Books, Huntington, USA, 2001.

Original ultrasound image, and two results with different amount of smoothing.

Following examples illustrate a low-complexity multiresolution method

A. Pizurica, W. Philips, I. Lemahieu and M. Acheroy, "A Versatile Wavelet Domain Noise Filtration Technique for Medical Imaging," IEEE, Trans. on Medical Imaging, vol. 22, no. 3, pp. 323--331, March 2003.

Gradual speckle suppression

A tunable parameter T exists. The final choice here is up to the medical expert.

Left to right: original ultrasound image, and two results with different amount of smoothing (T=2 and T=4).

Left to right: original ultrasound image, and two results with different amount of smoothing (T=2 and T=4).

Automatic processing

The parameter value can be fixed to the value that maximizes the signal to noise ratio. The method operates then fully automatically.

Left: original. Middle: homomorphic Wiener. Right: the proposed method.

Results with synthetic speckle

In this example, the test image is artifficially corrupted by spatially correlated synthetic speckle.

Left to right: noise-free image, artificial speckle (sigma=0.9), the results of the homomorphic Wiener and the proposed filter.