||Investigation of Ultrasound Vector Flow Imaging on a Commercial Platform
||Pihl, Michael Johannes
||Jensen, Jørgen Arendt (Biomedical Engineering, Department of Electrical Engineering, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
||Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
||Estimation of blood flow velocities plays a key role in diagnosing major diseases in e.g. the
carotid arteries. However, the velocities estimated using conventional Colour Flow Mapping or
Doppler techniques are angle dependent. The Transverse Oscillation (TO) method solves this
angle dependency and is able to estimate the transverse velocity component. The TO method
is based on a creation of a double oscillating field by using special apodization profiles when
beamforming in receive.
Previously reported results have been obtained with the experimental scanner RASMUS. To
investigate the feasibility of a commercial implementation, the performance of the TO method
is investigated based on beamformed radio frequency data from a commercial scanner. The
commercial scanner is set up in CFM mode with only minor modifications.
The TO estimator has been implemented in C, for increased speed in calculations and for a potential
later direct implementation in a scanner. Measurements were performed on a circulating
flow rig in order to measure the performance of the estimator and the setup. The estimated
transverse profile was compared to the theoretical parabolic flow profile, and the relative mean
standard deviation, ~¾, and relative mean bias, ~B , were calculated.
A number of parameters were change: Averaging length in the estimator, the number of shots
per estimate, transmit focal depth, beam to flow angle, number of transmit cycles, centre frequency,
and the transverse lag. The initial setup yielded for the transverse velocity ~¾vx=15.9%
and ~Bvx=17.4%. Using a centre frequency of 5 MHz and increasing the number of shots per
estimate to 64 reduced the relative standard deviation to 5.4%, whereas the bias was unaffected.
At a beam to flow angle of 90± ~¾vxwas 11.4% and ~Bvxwas 20.9%. This is in strong contrast to
a conventional velocity estimator. The conventional velocity estimator is not eable to estimate
any velocity of significant magnitude. This illustrates the major improvement of the velocity
vector estimation the TO methods provides.
The results were compared to previously reported simulations and flow rig measurements using
the experimental scanner RASMUS by Udesen et al. The results were comparable, with both
simulations and flow rig measurements. Adjusting for the differences in parameter values, the
relative standard deviations obtained were close to the previously reported values. The bias,
however, was somewhat higher. That is due to limitations in the apodization setup. With
suggested procedures and optimisation of the setup, this bias can be reduced. Simulations
should be performed in order to optimise the double-oscillating pulse-echo field.
The present study demonstrates the feasibility of a commercial implementation of the Transverse
Oscillation method for real-time vector velocity estimation with expected clinical impact
on disease diagnosis and potential benefits in for instance stroke prevention.
||Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Creation date: 2009-06-18
Update date: 2009-11-04