Two-dimensional image reconstruction with spectrally-randomized ultrasound signals

Abstract

An ultrasound imaging method using unfocused frequency-randomized transmissions and image reconstruction from data received by a single element is experimentally demonstrated. The elements of an ultrasound imaging array are randomly assigned different frequencies and driven by a multicycle sinusoidal burst. The resulting acoustic field is spectrally unique and target localization is possible based on the a priori knowledge of this field. A 64-element phased array driven by arbitrary waveform generators is used in the experiments. Transmission frequencies range from 2.00 to 2.64 MHz with 10 kHz resolution. One element of the array is reserved for receiving backscattered signals and an image is reconstructed from the signals received by this single element. Reconstruction is based on cross-correlation of the received data with transmitted bursts to obtain radial elliptical projections. Multiple projections are obtained from single received data, which are back-projected to obtain an image. Successful target localization is made possible through multiple frequency-randomized acquisitions. The performance of the method is measured using images of a single point target. These images are quantified and analyzed in terms of their point spread function (PSF) and SNR. Optimum imaging parameters, such as the number of acquisitions, transmit burst length, and number of possible receivers, are obtained through further analysis of SNR. Images obtained with the frequency-randomized transmission method compared well with the performance measurements of a typical B-mode acquisition. It is demonstrated that the frequency-randomized method provides images superior to B-mode images in terms of PSF. The two-point discrimination threshold is measured to be 2 mm in the lateral and azimuth directions.