On perfusion estimation using ultrasound

Research output: ThesisDoctoral Thesis (compilation)


This thesis has the main focus on ways to improve the blood perfusion estimate, as obtained from a continuous wave Doppler system. Earlier studies have shown large temporal variations in the detected perfusion estimate, due to speckle. Here, a theoretical model to predict frequency dependent characteristics of speckle in continuous wave ultrasound is presented. The model predicts the frequency shift which is necessary to obtain uncorrelated power in a continuous wave system insonating a medium with a random set of (static) scatterers. This is done by deriving a covariance function for received power, which functionally is the square of the (deterministic) autocorrelation function of the sensitivity function for the system, defined as the product of the transmit and receive beams. The model is valid in the far field and under the Born approximation. It is verified experimentally, using an agar phantom containing randomly dispersed scatterers.

Motion artifacts are shown to be reduced using a subtraction procedure. If frequencies are chosen in the multi-frequency system so that different channels produce uncorrelated signals from blood, using the theory described above, signals stemming from moving tissue are still correlated. By subtracting perfusion estimates, the correlated part can be suppressed. The method is shown both theoretically and experimentally to produce linear estimates of perfusion. It is also shown that the decorrelation of the squared amplitude of the Doppler phasor, versus transmitted frequency, closely follow what is predicted for the static case. Artifacts are suppressed a factor 2 to 4, but filtering of the signals prior to subtraction can improve the suppression substantially. The immediate drawback is then that the perfusion estimate becomes nonlinear.

The theoretical framework for the work mentioned above, is based on a general solution for the received signal in a two transducer system, when a statistically homogeneous medium is insonated. The theory is applied from studies where intrinsic scattering object properties are investigated. This theory and a unique ring transducer where such measurements can be performed, are reviewed. A procedure to obtain intrinsic scattering object properties is also reported. Measurements of the scattered acoustic field as a function of angle and frequency were normalized for system effects associated with the ring transducer system. The measurements yield the average differential scattering cross section, which under the Born approximation is directly proportional to the spatial-frequency spectrum of the medium inhomogeneities. Measured results for two phantoms consisting of glass microspheres embedded in agar show good relative agreement to theoretical cross sections for distributions of glass spheres measured experimentally.

Finally, a method named Ultrasound Doppler vector tomography is presented that reconstructs images of two-dimensional flow fields using a tomographic algorithm. Continuous wave Doppler measurements obtained in a plane from points encircling the region of interest, give measurement data suited for fan beam tomography. The reconstruction recovers the curl of the velocity field, and images based on measurement data from two different flow phantoms show good conformity to simulated results.
Original languageEnglish
Awarding Institution
  • Department of Biomedical Engineering
  • [unknown], [unknown], Supervisor, External person
Award date1999 Nov 26
Publication statusPublished - 1999

Bibliographical note

Defence details

Date: 1999-11-26
Time: 10:15
Place: Lund Institute of Technology, Ole Römers Väg 3, room E:1406

External reviewer(s)

Name: Burns, Peter N.
Title: Prof.
Affiliation: University of Toronto, Toronto, Canada


Subject classification (UKÄ)

  • Medical Engineering

Free keywords

  • Instrumentation technology
  • Speckle
  • Artifacts
  • Perfusion
  • Ultrasound
  • Doppler
  • Mät- och instrumenteringsteknik


Dive into the research topics of 'On perfusion estimation using ultrasound'. Together they form a unique fingerprint.

Cite this