On the Interpolation of Echographic Images Exploiting Band-Pass Signals

This paper deals with the study of an innovative interpolation technique aimed at the generation of echographic images of a synthetic scene previously insonified. The traditional echographic equipments perform the interpolation process working on the data obtained after the envelope extraction (low- pass data). The proposed method carries out the interpolation process using the band-pass data before the envelope extraction, performing, in this way, a complex values interpolation. I. INTRODUCTION This work refers to the ultrasound imaging. Among all of the possible applications, here we consider the one referring to the generation of medical echographic images. The purposes of this work is the development of new band-pass interpolation techniques aimed to the composition of the echographic images, comparing the results with those obtained adopting the low-pass interpolation methods nowadays in use. The traditional echographic equipments perform the interpolation process on the image pixels, working, in this way, on the gray levels obtained extracting the envelope of the demodulated echoes received by the array and processed by the beamforming algorithm. The linear interpolation function is commonly adopted because it is less burdensome in terms of computational load and provides better results when the spatial sampling theorem is respected. Otherwise, the designed method interpolates directly the band-pass signals obtained by the beamforming process. It is useful to interpolate before the envelope extraction process, since this process causes loss of phase information, which is important to obtain a better image, especially in condition of strong undersampling. In particular, in this paper, we try to demonstrate that the interpolation process applied to the band-pass signals instead of the low-pass signals (after envelope extraction) allows to obtain better results in terms of reconstructed images quality also, as mentioned above, when the spatial sampling theorem is not respected. This characteristic makes it possible to acquire the echo signals in undersampling condition, and thus, reconstructing a fully spatial sampled image using the addressed interpolation method. Some tests regarding the adoption of interpolation functions more complex than the linear one have been carried out on both band-pass and low-pass signals. Even though the implementation of these functions produces an unavoidable computational load growth, the quality of the obtained images is better than those interpolated with traditionally linear functions. In particular, our attention has been