Ultrasonic reflection tomography: Signal processing and resolution

Abstract

Medical imaging is, with nondestructive testing of materials, the main application of ultrasonic reflection tomography (URT). This method results from an inverse born approximation (IBA). URT allows perturbations (theoretically small) of a reference medium to be visualized. One is then concerned with a Fourier synthesis problem and the first step is to correctly cover the frequency space of the object by an adequate spatial and frequency scanning. In practice, the frequency bandwidth of the used ultrasonic pulses is limited by the transducers, leading to a poor coverage of the object’s spectrum and to a relatively poor resolution of images. Low and high frequencies are restored by a Papoulis deconvolution method that presents advantage not to modify the spectral content of the signal within its frequency band. The resolution enhancement procedure was first tested on academic targets (wire), and on test targets (square aluminum and triangular PVC rods), showing the possibility to obtain high-resolution images with low-frequency transducers. Finally, we compare the resolution given by a 250-kHz transducer to the one obtained, with a classical technique, by a 2,5 MHz transducer. The overall dimension of test objects is in that case less than the wavelength.