Real Time Delay Estimation and Correction for a new dual Frequency Method in Ultrasound Imaging

Abstract

A new dual frequency ultrasound imaging technique, uses a low frequency pulse to manipulate the propagation velocity and scattering conditions for the high frequency imaging pulse. With this new technique, the non-linearly scattered signal and new tissue parameters can be imaged. Signals transmitted under different manipulation pressures will be delayed with the respect to each other, and the delay will be varying with depth. This delay must be estimated and the signals corrected with the estimated delay, to be able to extract the non-linear scattering. In this thesis several different methods for solving the time delay estimation and correction problem is investigated. The methods are evaluated according to their numerical and computational performance. For time delay estimation several new estimators are proposed. The proposed WLS method is shown to perform best according to the measures used here. For delay correction standard linear interpolation was chosen. Current experimental two frequency probes are annular arrays which scans the image plane mechanically, transmitting one pulse in each beam direction, switching polarity of the low frequency pulse for each beam. Since two beams, one positively and one negatively manipulated, are needed in each direction an lateral upsampling is needed. Different upsampling techniques are also studied, and evaluated in the same way as the delay estimation and correction methods. The chosen methods are implemented for processing on standard CPUs and GPUs(Graphical Processing Unit). Runtime experiments show that the processing can be carried out in real time in both CPU and GPU. Frame rates up to 100 was achieved, using hardware found in standard desktop PCs.