Deconvolution of freehand 3d ultrasound data using improved reconstruction techniques in consideration of ultrasound point spread functions

Abstract

Three dimensional ultrasound volume datasets can be computed by volume reconstruction based on a static scan conversion using matrix array transducers or dynamic volume reconstruction based on the position and angle data of several single ultrasound data slices of a tracked linear array ultrasound transducer. The fact that medical ultrasound data suffers from blur caused by the volume expansion of the pressure field of the mechanical wave leads to signal deterioration that is dependent on the used excitation pulse and focusing of the ultrasonic wave. These sources can be examined in order to improve the overall system resolution for 3D ultrasound reconstructions of freehand tracked measurements by compensating it using deconvolution techniques or multicode compounding during the volume reconstruction step. In contrast to the static volume reconstruction of matrix array volume data freehand recorded data slices have no predictable positions and angles so that a volume reconstruction with regard to a blur compensation is a complex computation. Looking at the ultrasound transfer function we can focus on the simulation and measurement of the “point spread function” especially in the lateral and elevational direction. To understand its effects on a 3D reconstruction we compute a simulation of freehand-ultrasound slices based on synthetic phantom structures and given US parameters. Computing a 3D reconstruction of these simulated slices we can optimize the reconstruction algorithm itself to archive better resolution in the volume data sets considering ultrasound parameters like beamforming and the excitation pulses.