University of Michigan Banner
  Basic Radiological Sciences Ultrasound Group 
  Home Map Links People Contact
 
 

3D Flow Vector Generation From Decorrelation


TW Ketai, TA Tuthill, JB Fowlkes


Background/Objective

The purpose of this research is to develop a new angle independent technique for measuring blood flow speed and direction using ultrasound. It combines the established technique of Doppler imaging to determine in-plane velocities and decorrelation to determine out-of-plane velocities. These three components can then be combined to determine total 3D velocities. Decorrelation, D, represents the change in pixel intensity over time as the image changes, and can be related to velocity, V, by the formula D^2=Vx^2/Bx^2+Vy^2/By^2+Vz^2/Bz^2 where Bx, By and Bz are properties of the transducer's beamwidth in the axial, lateral and elevational directions.


Methods

For this research, a set of experiments involving the simulation of blood vessels in the body was conducted. A saline solution with contrast agent was chosen as a suitable blood-mimicking fluid. The experimental setup consisted of a tank filled with water with 6.4 mm dialysis tubing running through it. A 10 MHz transducer was positioned above the tube first perpendicular, then rotated to 20 and 40 degrees from perpendicular. As the saline and contrast agent was pumped through the tube at 15 ml/min, cine loops were collected with the beam steered 20 degrees to the left and right. These data were then processed using MATLAB programs to determine the direction and velocity of the flow.


Data/Results

The processing of the data we obtained produced the three vectors plots shown below. They may be rotated to visualize the slant of the vectors by playing the movies or by sliding the marker on the play bar.
 

The vectors above were generated from an experiment where the fluid was flowing normal (at 90 degrees) to the plane of the beam.


In the second trial, the fluid was flowing at 70 degrees laterally to the beam.


Here, the flow was at 50 degrees to the beam. The displayed vectors are not at 50 degrees from horizontal because of aliasing that occurred due to the limited frame rate.


Discussion

In current clinical settings, to measure blood velocities using Doppler, the sonographer must estimate the angle between the transducer and vessel. Our technique automates this process for determining the vessel orientation. The data above show that the method for finding and displaying the velocity vectors works fairly well. The error in the angle estimate for trial 3 is due to the aliasing in the Doppler measurements. The large velocity component in the lateral (x) direction results in a large frequency shift that can not be measured with the limited frame rate. With a higher frame rate or use of a higher firing rate along a single line, this error would be eliminated. For the smaller angles, however the data show that our method can, using grayscale images and doppler data, determine the angle at which a fluid in the field is flowing.


Conclusions

In conclusion, this method for flow analysis looks very promising. We need to do more trials to determine the accuracy to which the techinique works, but if this comes out to a reasonable range, this could be very helpful in medical procedures where the details of blood flow in the patient are needed. By scanning a vessel twice from the same location, with the beam of teh transducer steered oppositely in each scan, it should be possible to determine how fast and in what direction the blood in the vessel is flowing.
 
 
 
With the occasional exception of embedded movies, this site is HTML 3.2 compliant.
Send comments to GL LeCarpentier gllec@umich.edu (see listing)
or see other personnel listings.