Evaluation of volume vascularization index and flow index: a phantom study

Authors


Abstract

Objectives

Three-dimensional (3D) power Doppler ultrasonography provides indices to quantify moving blood within a volume of interest (e.g. ovary, endometrium, tumor or placenta). The purpose of this study was to determine the influence of ultrasound instrument settings on vascularization index (VI) and flow index (FI) at different flow velocities, using a specially built flow phantom with a small tube diameter.

Methods

Blood-mimicking fluid was pumped at 10–100 mL/h through a plastic tube with a diameter of 0.65 mm within a virtual spherical volume (content 137.12 cm3) of a Voluson 730 Expert 3D power Doppler ultrasound instrument. VI and FI were determined at different pulse repetition frequency (PRF) settings, with minimal and maximal wall motion filter (WMF) settings. The measured VI was compared with the actual VI.

Results

The ability to measure VI and FI at different flow velocities was highly dependent on the PRF and WMF settings. In our experimental set-up, using a PRF of 0.3 kHz, flow velocities of about 2 cm/s and higher could be registered. Measured VI was overestimated up to 44 times relative to actual VI.

Conclusions

Our main finding in a laboratory set-up was a considerable overestimation of moving blood volume using 3D power Doppler ultrasound in a single small tube. The degree of overestimation depends on the spatial resolution and on the settings of the ultrasound instrument. When small vessels are involved in a clinical setting, interpretation of VI should take this overestimation of moving blood volume into account. Copyright © 2008 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Frequency-based color Doppler diagnostics can be extended by the use of power Doppler sonography, especially in situations of low-velocity blood flow1. Power Doppler is more sensitive in terms of flow detection, less dependent on insonation angle and not susceptible to aliasing2. During standard color Doppler sonography the mean frequency shift of the back-scattered signals is analyzed to give information on velocity and direction. Power Doppler, on the other hand, uses a color map to encode the integrated power from the Doppler signal, with the value of color-coded pixels relating to the concentration of scatterers, i.e. detectably moving blood cells3.

The combined use of power Doppler with three-dimensional (3D) ultrasound provides the possibility of quantifying moving blood within a volume of interest4. Three indices are calculated, namely vascularization index (VI), flow index (FI) and vascularization flow index (VFI)4–6. VI, which is expressed as a percentage, denotes the ratio of color-coded voxels to all voxels in a given volume of interest. This is an estimate of the percentage of the volume filled with detectably moving blood. FI refers to the mean value of all color-coded voxels in the vessels of the volume analyzed. The value of each color-coded voxel is expressed by the ultrasound instrument in arbitrary units on a scale of 0 to 100. Multiplying VI by FI derives VFI.

3D power Doppler ultrasonography can be applied in all fields of medicine where quantification of vascularization, neovascularization or revascularization is of potential importance in clinical decision making. Unfortunately, the quantity one wishes to measure, namely the density of blood vessels, is only one of the factors that affect the color-coded voxel count7. Other important factors include blood-flow velocity and ultrasound instrument settings7–10. Furthermore, 3D power Doppler ultrasound can be applied to quantify vessel stenosis. In a stenotic vessel phantom simulating carotid or femoral artery stenosis, quantification of relatively large vessel areas (10–50 mm2) has been shown to be fairly accurate (overall accuracy 8.3% of the actual vessel area)11. However, vascularization also involves microvascular networks, including small arteries, small veins and capillaries. To the best of our knowledge, data on in-vitro quantification of such smaller vessel areas using 3D power Doppler are not yet available in the literature. The purpose of this study was to determine the influence of ultrasound instrument settings on VI and FI, using a tissue-equivalent phantom and a specially built flow phantom with a small tube diameter at different flow velocities.

Methods

A Voluson 730 Expert 3D ultrasound instrument (Kretz, GE Medical Systems, Zipf, Austria) was used with a mechanical 4.3–7.5-MHz vaginal transducer probe (RIC5-9) and two types of phantom.

  • 1)Tissue-equivalent phantom: The influence of receiver gain settings of the ultrasound instrument on VI and FI was assessed in a multi-purpose tissue-equivalent phantom (RMI Cat # 84–317, Radiation Measurements Inc., Middleton, USA). Using a mechanical holder and applying commercially available coupling gel the probe was attached to the tissue phantom.
  • 2)Flow phantom: A custom-made flow phantom was built in collaboration with the Department of Experimental Cardiology at the Erasmus Medical Center (Figure 1). A plexiglas box was filled with a mixture of water and coupling gel in order to minimize vibrations. Into this liquid a holding apparatus was put, containing a holder to guide a polyurethane test tube with a length of 21 cm and an inner diameter of 0.65 mm. The angle of the tube with the liquid surface was 45°. Flow was generated with a Hospal K-10 infusion pump (Medolla, Italy) in 10 mL/h steps from 0 up to a maximum of 100 mL/h. A flow of 100 mL/h in a tube with an inner diameter of 0.65 mm results in a mean flow velocity of 8.37 cm/s. This flow velocity corresponds to the maximum blood-flow velocity in small vessels of similar size in vivo12. A blood-mimicking Doppler fluid for Doppler flow phantoms (Dansk Fantom Service, Jyllinge, Denmark) was used.
Figure 1.

Photograph of custom-made flow phantom.

Fixed to a mechanical holder, the tip of the probe was immersed in the liquid just above the tube. The following constant default instrument settings (corresponding to the manufacturer's first-trimester settings for power Doppler) were used throughout the experiments: frequency, mid, 6.0 MHz; dynamic, set 3; balance, > 150; smooth, 5/5; ensemble, 15; line density, 9; power Doppler map, 5; artifact suppression, off; power Doppler line filter, off; quality, high. After obtaining a volume scan in 3D power Doppler mode, the saved data were processed as follows. A part of the tube with a length of 64 mm was marked with lengths of tape that were clearly visible on ultrasonography. Care was taken to affix the tape in a manner such that the tube would not be constricted at the marking points. Between these two marking points, using the integrated Virtual Organ Computer-aided AnaLysis (VOCAL) imaging software, a virtual spherical volume with a content of 137.12 cm3 was constructed. Moving blood volume was simulated by the blood-mimicking fluid flow through the tube. VI and FI were determined using the volume histogram facility and published formulae5, 6. The measurements were repeated using different pulse repetition frequency (PRF) settings, in combination with both minimal and maximal wall motion filter (WMF) settings on the ultrasound instrument. The actual VI was calculated as the ratio of the tube volume (0.021 cm3) to the virtual spherical volume, expressed as a percentage (0.015%), and the measured VI was compared with the actual VI.

Results

Tissue-equivalent phantom

Since there is no flow in this phantom, detection of any flow must originate from the amplification of instrument noise. The findings of the tissue-phantom experiment are shown in Figures 2 and 3. Flow—expressed as VI or FI—was not detected below a gain setting of − 7. Higher gain settings resulted in false flow detection. From a gain setting of − 2 or − 1 onwards, VI and FI showed sharp increases up to 98% and 89 arbitrary units, respectively, at still higher gain settings. The lowest detectable FI value was about 20 arbitrary units. These characteristics were not different using the minimal (0.1 kHz) or maximal (7.5 kHz) PRF on the Voluson 730 Expert. Based on these results it was decided to use a gain of − 7 in the flow-phantom experiment.

Figure 2.

Vascularization index plotted against gain in a flow-free tissue-equivalent phantom. Pulse repetition frequency of 0.1 kHz (●) and 7.5 kHz (○).

Figure 3.

Flow index plotted against gain in a flow-free tissue-equivalent phantom. Pulse repetition frequency of 0.1 kHz (●) and 7.5 kHz (○).

Flow phantom

Using the lowest possible PRF (0.1 kHz) resulted in the detection of slow-moving particles in the liquid in which the phantom was immersed. Therefore we decided to skip this lowest PRF setting in further measurements. Figures 4 and 5 show the relationship between VI and FI and actually generated mean flow velocities, as measured at different PRF and/or WMF settings. The mean flow velocity (cm/s) was calculated by dividing the pre-installed volume flow by the tube's cross-sectional area.

Figure 4.

Vascularization index (VI) plotted against mean flow velocity in a flow phantom. Pulse repetition frequency (PRF) of 0.3 kHz, with minimal (●) and maximal (○) wall motion filter (WMF); PRF of 2.4 kHz with minimal (▪) and maximal (□) WMF; PRF of 5 kHz with minimal (▴) and maximal (▵) WMF. The actual VI (0.015%) is plotted as a horizontal line.

Figure 5.

Flow index plotted against mean flow velocity in a flow phantom. Pulse repetition frequency (PRF) of 0.3 kHz, with minimal (●) and maximal (○) wall motion filter (WMF); PRF of 2.4 kHz with minimal (▪) and maximal (□) WMF; PRF of 5 kHz with minimal (▴) and maximal (▵) WMF.

Figure 4 shows that by changing the lowest generated flow velocity of 0.84 cm/s to the next velocity step of 1.68 cm/s, using a PRF of 0.3 kHz with minimal WMF, VI increased from a value of 0.209% to a value of 0.541% around which, at higher flow velocities, it stabilized. Using the same PRF and an identical velocity step, but in combination with maximal WMF, VI increased from a value of 0.087% to a value of 0.445% around which, at higher flow velocities, it stabilized. Using a PRF of 2.4 kHz with minimal and maximal WMF, VI detection started at flow velocities of 2.52 cm/s and 5.04 cm/s, respectively. VI increased at higher flow velocities, to level off around values of 0.549% and 0.403%, respectively, at flow velocities of 7.56 cm/s and higher. Using a PRF of 5 kHz with minimal WMF, VI detection started at a flow velocity of 5.86 cm/s, gradually increasing to a value of 0.215% at the highest generated flow velocity of 8.37 cm/s. Using a PRF of 5 kHz with maximal WMF, all VI values were zero. The actual VI is represented in Figure 4 as a horizontal line at 0.015% on the y-axis. All detectable 6 values were overestimated relative to this actual VI, up to approximately 44 times, using a PRF of 0.3 and 2.4 kHz, both with minimal WMF.

Figure 5 shows the lowest detectable FI value to be about 20 arbitrary units, regardless of PRF and WMF settings. Using a PRF of 0.3 kHz with minimal and maximal WMF, at flow velocities of 2.52 cm/s and higher and using a PRF of 2.4 kHz with minimal WMF at flow velocities of 5.86 cm/s and higher, FI stabilized around a value of 36 arbitrary units. Using a PRF of 2.4 kHz with maximal WMF and using a PRF of 5 kHz with minimal WMF, FI did not stabilize up to and including the highest generated flow velocity of 8.37 cm/s. Using a PRF of 5 kHz with maximal WMF, all FI values were around the lowest detectable value.

Discussion

Ever since its introduction, there have been hopes that the judicious use of power Doppler ultrasonography would be of great value in flow diagnostics, because of its ability to detect blood flow at low velocities13. Using power Doppler in combination with 3D ultrasound to reliably quantify vascularization in a given volume of tissue, could be of much clinical interest in obstetrics and gynecology, for the evaluation of reproductive processes14, tumor angiogenesis15 or placental vascularity16, to name but a few indications.

It is well known to ultrasound operators that turning up the gain function while power Doppler mode is switched on will eventually produce color signals, even when the transducer is still in its holder. In the flow-free tissue-equivalent phantom in this study we attempted to clarify from which gain setting onwards color information is given, while there is no actual flow. Using the Voluson 730 Expert, gain settings higher than a quarter of the full range, i.e. a gain of more than − 7, resulted in false flow detection. We cannot explain why VI and FI values slightly decreased at gain 5 or more. However, in daily practice these extremely high gain settings will not be used, as this will result in clearly recognizable false color signals. In general, for any power Doppler instrument, to avoid false flow detection, we advise assessment of the proper gain setting. Strictly speaking, a complete flow-free area in living tissue does not exist. Therefore, ideally, a tissue-equivalent phantom should be used for this assessment.

High-pass filtering is necessary to discriminate vessel-wall motion from blood-cell movement. Therefore, the ability to detect low flow velocities using power Doppler is limited. As a consequence, VI and FI in 3D power Doppler mode are influenced by WMF settings on the ultrasound instrument, as shown by our flow-phantom experiment. The velocity-dependent increase for VI and FI over a limited range is an effect of the WMF. This same finding was reported in a phantom experiment by Jansson et al.17. Moreover, since the manufacturers of the Voluson 730 Expert have chosen to set the WMF as a percentage of the PRF (low–median–high WMF setting refers to an increasing fraction of PRF), VI and FI are influenced by PRF settings as well. Using higher PRF and/or higher WMF settings, the lowest flow velocities are not detected.

Based on our results, using the Voluson 730 Expert in order to detect very low flow velocities in a clinical setting, we advise the use of the lowest possible PRF (of at least 0.3 kHz) with minimal WMF setting. In this way, we found a stabilization of VI and FI at flow velocities of 1.68 cm/s and higher. When extrapolating this finding to the human capillary system in vivo, where the flow rate is no more than a few mm/s12, it is clear that moving blood volume within this part of the circulation is not detected. Furthermore, detectable flow velocities in the range of 1 to 5 cm/s are found in small arteries and veins. However, as the lumen diameter of these vessels is no greater than approximately 0.5 mm, as demonstrated in this study, an overestimation of moving blood volume up to 44 times might be expected. The degree of overestimation depends on the spatial resolution and on the settings of the ultrasound instrument. Thus at this time, 3D power Doppler ultrasound cannot be used for true measurements of moving blood volume when small vessels are involved. However, VI may prove its value for assessing trends or relative changes in vascularity.

FI values less than 20 arbitrary units were not detected, indicating that the Voluson 730 Expert uses a threshold setting to discriminate between true velocities and noise. In addition, depth dependency of FI has been demonstrated18. Outside the attenuation region of the WMF, FI is not related to flow. Therefore the term ‘flow index’ does not seem to be appropriate. Instead, as FI reflects the mean power of the Doppler signal, the term ‘power index’ may be preferable.

In conclusion, it is now possible—using 3D power Doppler ultrasound and digital imaging techniques—to quantify moving blood in vivo, albeit with arbitrary indices. Care must be taken in comparison and interpretation of the numerical information generated. If VI and FI are to be compared, instrument settings must be kept constant. 3D power Doppler ultrasound cannot be used for the true measurement of vascularity, and the clinical value of VI and FI remains to be established.

Ancillary