Real-Time Contrast-Free Ultrasonic Blood Flow

Velocity Profile Measurement

G.G. Koutsouridis, N. Bijnens, P.J. Brands, F.N. van de Vosse and M.C.M. Rutten

Problem Description1. Ultrasonic Perpendicular Velocimetry Wall

Doppler Ultrasound UPVFigure 1 Current ultrasound research system with real-time GPU module design for the UPV method

1. Ultrasonic Perpendicular Velocimetry

(UPV) to RF-data for accurate

velocity and flow assessment is time

consuming due to data size and post

processing [1,2].

WallUltrasound beam Ultrasound beam

Centerline

processing [1,2].

2. Fast Fourier Transform (FFT) on

Butterworth Band Pass Filters (BPF)

for vessel’s wall removal requires

contrast agents dispersion in the fluid

for the application of Cross-

correlation.

Aim In-vitro (aortic-like polyurethane vessel) with Blood Mimicking Fluid (BMF), resembling the

Methods

AimReal-time UPV on Graphics

Processing Unit (GPU) [3].

In-vitro (aortic-like polyurethane vessel) with Blood Mimicking Fluid (BMF), resembling the

rheological (shear thinning) & acoustical (backscattering) Blood’s properties, as contrast agent.

Ex-vivo (porcine carotid arteries) with BMF & contrast-free real Blood, implementing Wavelet

Transform (WT) filtering.

ResultsMethods• In-vitro and Ex-vivo constant and physiological pulsating

flows of 1Hz and peak flow velocity 0.8m/sec.

• Ultrasound RF-data acquired at 33MHz (Fast B-mode)

� 6 seconds of images, at 730frames/sec.

ResultsAcceleration offered by the UPV method for the real-time

(Figure 3.[#]) perpendicular assessment of the velocity profiles.

� 6 seconds of images, at 730frames/sec.

For the UPV technique, the process consisted, among

others (Figure 1), of the following steps:

� Wall removal and contrast-free fluid scattering

enhancement via WT Daubechies8 filtering (localization

3.[1] 3.[2]

enhancement via WT Daubechies8 filtering (localization

in time & frequency) for functions with discontinuities and

sharp peaks (Figure 2).

� High Pass Filtering (HPF) of the sequential frames in the

time direction.

(i) (ii) (iii) (i) (ii) (iii)In-vitro constant flow BMF In-vitro pulsating flow BMF

� Application of improved Cross-correlation for the

assessment of velocity profiles.

3.[3]0.7

0.6

0.7

0.6

(i) (ii) (iii) Ex-vivo pulsating flow Blood 0.4

0.5

0.6

0

0.2

0.4

0.6

va

x(m

/s)

0.4

0.5

0.6

0

0.2

0.4

0.6

va

x(m

/s)

Figure 3.[#] Real-time GPU velocity profile (▪) assessment with: (i) No filtering, (ii) WT and (iii) WT & HPF

0.2

0.3

0.4

v(m

/s) -5 0 5

x 10-3r(m)

0.04

0.06

v (

m/s

)

0.2

0.3

0.4

v(m

/s) -5 0 5

x 10-3

r(m)

0.04

0.06

v (

m/s

)

ConclusionThe WT filtering technique allows measurement of Blood flow

without contrast agents, while the GPU allows a real-time

-5 0 5-0.1

0

0.1

-5 0 50

0.02

0.04

∆ v

(m

/s)

-5 0 5-3

-0.1

0

0.1

r(m)

-5 0 5-3

0

0.02

0.04

∆ v

(m

/s)

without contrast agents, while the GPU allows a real-time

assessment of axial velocity distribution. With a real-time

implementation of the local ultrasound pressure estimation,

real-time characteristic vascular impedance assessment, as a

diagnostic tool, will be feasible even In-vivo.

Figure 2 The effect of WT Daub8 (right) vs BPF (left) on the quality of thevelocity profiles assessed by UPV. Left panels show estimates fromindividual frames, while top and bottom indicate mean values andstandard deviations

-5 0 5

x 10-3r(m)

-5 0 5

x 10-3r(m)x 10

-3r(m) x 10

-3

r(m)diagnostic tool, will be feasible even In-vivo.References[1] Beulen, B.W.A.M.M. et al. (2010), “Perpendicular ultrasound velocity measurement by 2D cross correlation of RF data. Part A:

validation in a straight tube”, Exp Fluids, 49, pp. 1177-1186.

[2] Beulen, B.W.A.M.M. et al. (2011), “Toward noninvasive blood pressure assessment in arteries by using ultrasound”, Ultrasound in

Med. & Biol., 37:5, pp. 788-797.

[3] Owens, J.D. et al. (2008), “GPU Computing: Graphics Processing Units–powerful, programmable and highly parallel–are

increasingly targeting general-purpose computing applications”, Proceeding of the IEEE, 96:5, pp. 879-899.

/ Department of Biomedical Engineering