Deconvolution of medical ultrasound images via parametric inverse filtering

The finite frequency bandwidth of ultrasound transducers and the non-negligible width of transmitted acoustic beams are the most significant factors that limit the resolution of medical ultrasound imaging. As a result, in order to recover diagnostically important image details, which are often obscured due to the resolution limitations, an image restoration procedure should be applied. The current study addresses the problem of reconstructing ultrasound images by means of the blind deconvolution techniques. Particularly, the proposed deconvolution method is based on inversely filtering the complex-valued ultrasound images with a restoration kernel, whose Fourier transform is modeled as a member of a finite-dimensional, principal shift-invariant subspace. This approach presents a novel and very versatile way of modeling the frequency response of the inverse filter, in which the latter is defined by a few parameters, which can be estimated from the data using some reasonable assumptions on statistical properties of the tissue reflectivity. The effectiveness of the proposed method is demonstrated through a number of in silico and in vivo examples