Duplex and Color Doppler Imaging of the Venous System

The human venous system has, in the past, not alwaysattracted the same attention and interest from radiologists as the arterial system. Nevertheless, some of the pathologic conditions of the venous system are either frequently seen in daily practice such as varicosis of the lower extremity, or may relate to more severe and even life-threatening conditions,for instance deep venous thrombosis or liver venous pathology. However,enormous progress has been achieved during the past decade in the field of duplex and color Doppler imaging necessitating an update in our knowledge of these techniques. For both reasons, a volume specifically dealing with duplex and color Doppler ima­ ging of the venous system was considered an appropriate addition to the series Medical Radiology - Diagnostic Imaging. I would like to thank Prof. Mostbeck for his outstanding performance as the editor of this work. I would like to congratulate him and the many contributing authors, all renowned experts in the field, on their comprehensive coverage of the different topics, the up-to-date contents and the superb illustrations. I stronglybelievethat this bookwill encounter the same success as previous volumes published in this series.

• Language: English
• Publisher: Springer
• Release date: Dec 6, 2012
• ISBN: 9783642185892

Book preview

]>

1

Basic Principles and Physics of Duplex and Color Doppler Imaging

C. Kollmann PhD¹

(1)

Department of Biomedical Engineering & Physics, AKH Vienna, Währinger Gürtel 18-20, 1090, Vienna, Austria

CONTENTS

1.1 Duplex Systems

1.1.1 PW Doppler (Duplex System)

1.2 Color Doppler Imaging Based on Doppler Effect

1.2.1 Conventional Color Doppler Imaging

1.2.2 Amplitude-Coded Color Doppler Imaging (Power Doppler)

1.2.3 Harmonic Imaging

1.3 Color Doppler Imaging Independent of Doppler Effect

1.3.1 Time-Domain Analysis

1.4 Doppler Artifacts

1.5 Safety and Risks

1.5.1 Mechanical Index

1.5.2 Thermal Index

1.6 Quality Assurance of Doppler Systems

1.7 Ongoing Developments and Future Aspects in Color Doppler Imaging

References

Introduction

Doppler ultrasound has been used in medicine since the mid-1970s, mainly for the diagnosis of vascular disease like occlusion or stenosis (Keller et al. 1975; Fitzgerald et al. 1977; Weaver et al. 1980; Brown et al. 1982). The first systems developed were based on the continuous-wave (CW) spectral Doppler ultrasound technique with separate transmitter and receiver elements (Brody et al. 1974; Di Pietro et al. 1978). In 1983, pulsed-wave (PW) color Doppler imaging (CDI) systems were introduced clinically for color coding the detected blood flow in real-time (Atkinson et al. 1982). In the past, these early CDI systems were restricted to an evaluation of only a few well-defined medical indications in cardiac disease where the blood velocity is very high. A newer generation of CDI systems has allowed us, since 1986, to detect and display lower blood velocities occurring in peripheral arteries or veins (Kasai et al. 1985; Merritt et al. 1987; Klews 1987; Scoutt et al. 1990). The conventional CDI systems and the newer methods with their different signal processing algorithms (power Doppler, harmonic imaging) evaluate the blood velocity as a consequence of the Doppler effect (Fig. 1.1). Another method that has been clinically available since 1990 is the direct calculation of the blood velocity using a time domain analysis (Bonnefous et al. 1986; Klews 1991). With this technique, the change of the position of the moving red blood cells between two pulse intervals is detected and used for further signal processing steps (Fig. 1.1).

Fig. 1.1.Doppler imaging techniques and the physical principles for detecting blood flow

In contrast to CW Doppler systems, CDI systems were introduced to clinical routine applications rapidly, and the number of systems available in clinics was very large from the beginning (Schneider et al. 1993).

Familiarity with the basic principles of Doppler ultrasound is essential for its proper clinical use and an optimal diagnosis.

1.1 Duplex Systems

The simplest technique for detecting blood flow in real-time is a CW Doppler system consisting of two elements for transmitting the frequency signal continuously (ftransmit ≈ 2–10 MHz) and receiving a frequency signal (freceive) modulated by the Doppler effect according to

(1)

where Δf is the Doppler shift frequency, v the velocity of the erythrocytes, c=1540 m/s, and α the angle between the incident ultrasound beam and the direction of the blood flow (Fig. 1.2a). In the transducer, the two narrow band elements are arranged in a way that both ultrasound beams overlap to provide a small Doppler sample volume. From this volume, the frequency signal of the receiving element is modulated by the moving red blood cells (RBC). This signal is amplified and fed into a demodulator that compares the detected frequency with that from an oscillator to derive a signal difference equal to the Doppler shift frequency Δf. Most demodulators employ a technique called phase quadrature detection, which has the capacity to distinguish between blood flow towards or away from the transducer corresponding to a higher or lower received frequency, respectively. This bi-directional demodulation produces two output signals that have a phase relationship determined by the direction of flow. After high-pass filtering, typically in the range 50–250 Hz to reduce low-frequency noise but also low velocities (Fig. 1.3), a stereo audio signal is produced that can be fed to the loudspeakers, headphones, or displayed graphically. The audio signal of each stereo channel varies depending on the selected Doppler angle, transmitted frequency, and RBC velocity and is normally below 20 KHz (Table 1.1).

Fig. 1.2a,b.The techniques to acquire flow information from CW and PW Doppler by using the Doppler effect

Fig. 1.3a–c. Influence of the sample gate size and the wall-filter settings for PW Doppler spectral analysis measuring small flow velocities. a The sample gate is chosen large enough to cover the whole diameter of the vessel, and the wall-filter frequency is minimized (small black gap between zero-line and signal). The spectrum appears correct with a broad velocity distribution. b The wall-filter frequency is changed to its maximum value. The parts of low velocities are not displayed any more in the spectrum (black gap). c Additional to b, the sample gate size is minimized. This leads to a weaker spectrum. Only the velocities of a special vessel region are detected, i.e., the velocity distribution scanned is not as wide as the whole vessel diameter

Table 1.1. Theoretically derived Doppler shift frequencies in selected arterial and venous vessels (Taylor et al. 1988; Hennerici and Neuerburg-Heusler 1999; Seitz and Kubale 1988; Philips 1989) for an incident 45o ultrasound beam and three different transmit frequencies

CW Doppler is a highly sensitive tool for detecting the weak signals preferred for the examinations of smaller vessels like supraorbital arteries and veins or for ophthalmic and transcranial applications. Also, higher velocities than with PW Doppler systems can be monitored, and the effect of aliasing does not occur. The disadvantage with this technique which has led to it being displaced in most clinical diagnostic applications by PW Doppler techniques is the missing depth information of the detected Doppler signal. Because CW Doppler transmits its signal continuously, no run-time information can be used to identify the depth of the origin in the sample volume from which the received signal came (Fig. 1.2a). A CW Doppler technique is therefore not very helpful if many vessels are present in the region of interest. But in the early days of ultrasound Doppler or in low-cost systems where no Duplex technique is available, i.e., to locate the origin of flow within the B-mode image, this CW technique was used to detect blood vessels in a simple and easy way.

1.1.1 PW Doppler (Duplex System)

CW and PW Doppler are equal in the manner of signal post-processing and displaying the spectral results but different in signal generation. In PW Doppler systems, pulsed ultrasound waves are emitted with the same active element that is used for detection (Fig. 1.2b). Therefore, the electronic controller of a PW Doppler system has to switch between transmission and receiving mode. After the first short pulse sequence has been emitted, the system switches to the receiving mode and detects the reflected ultrasound waves after a time T1 for a certain time interval Δt, then the electronics switches back to transmission mode and emits the next pulse sequence. The time T1 is assigned to the selected scanning depth (sample volume start position), i.e., twice the time the ultrasound pulse needs to hit the Doppler targets, while during ΔT all the echoes with Doppler information are detected within a specific sample volume size. The maximum possible Doppler pulse repetition frequency (PRFmax) of the system is hence dependent on the time period of T1 + ΔT. According to the Nyquist theorem, the maximum frequency that can be detected without aliasing, that is unambiguously, at this depth is

(2)

If this limit is not exceeded, a PW Doppler system can determine the position of the moving RBCs, and after quadrature signal processing of the echoes, the information about the direction and velocity of the blood flow can be given (Fig. 1.2b).

For all systems that are based on the Doppler effect, this quadrature detector is a key component in the signal processing chain. In this detector, the analog echo signals are mixed with a reference signal from an oscillator (Fig. 1.4) to get information about the flow direction. The final result of this process consists of two different signals, the in-put or cosine component (I-signal) and the quadrature or sine component of the signal (Q-signal) with respect to the reference signal. The I-signal includes the temporal velocity information of the RBC echoes coming towards the transducer, while the Q-signal includes the particle’s temporal velocity information flowing away from the transducer. The next signal processing steps are shown in Fig. 1.4. The I- and Q-signals are analog/digitally converted (A/D), and the low-velocity components are removed by a moving target indicator. This can be done in the simplest way by a high-pass filter (wall filter).

Fig. 1.4.Block diagram showing the modules used for deriving the phase shift information from detected signals in PW spectral Doppler systems

Afterwards, the velocity itself of both signal components must be determined. This can be done by phase analysis (Fig. 1.4) or by a discrete Fourier analysis (DFT) of the frequency shift signals (Fig. 1.5). In the first spectral PW Doppler systems, a phase detection algorithm was implemented. In the following, the signal processing of the I-component is considered to describe the function of this detector. The same process is carried out for the Q-component. In the phase detector, the signal (signalI) is compared with a reference signal (signalref). If two waves with the same amplitude but different frequency start with the same phase angle (Φ=0) at a specific time (Fig. 1.4), signalref with the higher frequency will propagate faster than signalI. The phase angle Φref is at a specific moment higher than ΦI. That means that the wave with the shorter wavelength (λref) reaches a particular phase earlier than the wave with the longer wavelength (λI). For this reason, it is possible to determine the phase shift (ΔΦ) and consequently the frequency shift (Δf) or the difference of wavelength (Δλ) by comparing either signal’s phases. Now the temporal distribution of the velocity of the RBCs within the sample volume is derived by using Eq. 1 before this information is stored and processed for a spectral display. This flow information in combination with an overlaid B-mode image is known as the spectral Duplex technique. For a rough estimate, the maximum detectable velocities of these systems are proportional to the ratio between PRFmax and ftransmit. The estimated RBC velocities of these systems are only calculated and displayed correctly if the correction of the Doppler angle (α, Fig. 1.2), that can be set with a graphical symbol manually beforehand in the system, takes place in the right way. The errors of the velocity values are relatively low as long as the angle is between 10° and 60°. Other angle settings can lead to severe over- or underestimation of the real velocity and hence, in special cases where an absolute velocity information is needed, to an incorrect diagnosis. For the determination of the pulsatility (PI) or resistance (RI) indices, the Doppler angle is not so critical (Merrit 1991; Haerten and Mück 1992).

Fig. 1.5.Block diagram showing the technique and the modules used for deriving the frequency shift information from detected signals of real-time blood flow mapping systems

1.2 Color Doppler Imaging Based on Doppler Effect

In modern PW Doppler systems, not only the spectral display of the detected flow characteristics is available but also a color presentation. This is known as color Doppler imaging and is often used as a first orientation, whether flow can be detected in the scanned area or not, before a more specialized diagnosis of the flow follows. The presence of flow, its direction, speed, and type (e.g., laminar or turbulent) is indicated in special color schemes with hues, saturations, and brightnesses.

This chapter gives a survey of the basic principles of the favorite methods that are used today in these systems. Not only the frequency shift information (conventional color Doppler) from the echoes can be post-processed for a color representation, but also the strength of the amplitude of this frequency shift (amplitude-coded color Doppler) or the second harmonic of this frequency shift (harmonic imaging, Fig. 1.1). With the last two new methods, Doppler ultrasound has particularly pushed forward in regions that could not be evaluated or displayed with the conventional method before: Low flows in capillaries as well as the vascularity of tumors can be detected now.

1.2.1 Conventional Color Doppler Imaging

In conventional CDI, the signal processing is equal to the spectral Doppler mode (Fig. 1.5). A PW Doppler system emits a relatively long pulse sequence along a scan-line that can be steered by the user. A single gate within the sample volume (Fig. 1.6) is used to acquire the Doppler frequency shift data. In CDI, several gates per scan-line and within the same sample volume are implemented (Fig. 1.6). Each of these gates is capable of detecting Doppler signals at a time. In CDI, many parallel channels are connected to the input of the quadrature detector, and a large number of Doppler signals have to be processed simultaneously to derive a velocity profile and the spatial velocity distribution across the vessel lumen. Typically, 16–32 gates with approximately 1 mm axial length are available in a modern CDI system (Burns 1987). This causes a timing problem because the ultrasound beam must be stationary for a short time period at each scan-line to collect the Doppler signals. If many parallel channels exist, real-time processing of flow data by discrete Fourier analysis cannot be managed. To obtain the Doppler flow information along a large number of scan-lines and in real time, another method must be used.

Fig. 1.6.Number and distribution of the sample volume gates for the different Doppler techniques. While the PW spectral Doppler has got only one sample gate, on modern CDI systems there are several gates per scan-line where the velocity is measured for a color-coded representation

The autocorrelation analyzer serves exactly this function (Fig. 1.5). After real-time quadrature detection, the phase information curve of the I- and Q-components from each scan-line is periodical. If the I/Q-components are transferred to an I/Q-diagram (Fig. 1.7), their signal amplitudes can be symbolized by a vector with length L and direction Φ. During different time points (T, 2T,..), these vectors L are derived. In the autocorrelator, the particular temporal vectors are now multiplied: The I- and Q-components of these signals are multiplied with components from the same line but from a previous pulse (time delay). For a first output of the autocorrelator, 4–16 vector multiplications are used. After this process, there is a resulting vector of length R and direction (Fig. 1.7). From the resulting average phase shift angle , the average Doppler frequency shift . f can be derived easily (as shown for duplex systems earlier). According to Eq. 1, this Doppler frequency shift is proportional to the average flow velocity . While the length of the vector R is not used for further signal processing, the size and the sign of are stored and give an estimation in the color image of the average speed and direction of the flow velocity. The standard deviations of particularly DFi in relation to are used in the signal processing chain as a unit for flow stability and can be considered an indicator for turbulence.

Fig. 1.7.Schematic view of an autocorrelation process. A detailed explanation of the autocorrelation is given in the section on Conventional Color Doppler Imaging

Because of this way of signal processing, the problem of clutter or motion artifacts is severe in such a CDI system. The echoes of the erythrocytes are 100–1000 times weaker than those of solid targets (e.g., vessel walls) that are stretched by the pulsatility effect and as a result are moving slowly. This effect can inhibit the detection of the weaker Doppler shift echoes. The digital filtering algorithms implemented in the moving target detector or indicator (MTI) of a CDI system to discriminate between blood and tissue are highly sophisticated and averaged over frames to eliminate these clutter artifacts.

In a final step, the color information is overlaid on the B-mode image, creating a real-time color flow mapping system.

The detection of blood flow and its presentation in color can be manually set up using special boxes, the color windows. These windows contain the scan-lines for the color representation process (Fig. 1.6), and their shape can be changed by electronically steering the ultrasound beam coming out of the transducer. Modern transducer technologies with electronic beam steering (phased-array), electronic switching beams (linear-array), or a combination of both techniques can do this well.

Depending on the width, height, and position of this window within the scanned B-mode region, the frame rate (FR), i.e., the number of refreshing cycles of the flow information displayed on the monitor per second, changes. The larger and deeper a color window is set, the lower is the frame rate of the CDI system normally:

(3)

where PRF is the pulse repetition frequency, SL is the number of scan-lines in the color window, and N the number of pulses per color line.

If low flows are going to be detected with a CDI system, the PRF is decreased, and as a consequence according to Eq. 3, the frame rate is low (only several Hz). This effect can be sometimes recognized as flickering of the color information update on the monitor.

Another technique that is used for flow signal processing is the maximum entropy method (MEM). It is an autoregressive (AR) technique for spectral density analysis and maximizes the uncertainty, or entropy, of a time series expressed as an autocorrelation sequence. In some newer Doppler systems, this AR processing technique is implemented parallel to the other analyzer modules (Fig. 1.5). Some advantages of the AR technique are its real-time processing method and under certain circumstances its detection of lower flow velocities than with the other CDI techniques (Sohn et al. 1996; Battle et al. 1997).

All CDI methods and their accurate velocity presentation are influenced, of course, by aliasing and the Doppler angle.

The color represents only a single Doppler parameter and does not describe the full Doppler frequency spectrum. However, a spectral analysis (with PW Doppler) has more potential for a diagnosis than the pure color mode and should be preferred.

1.2.2 Amplitude-Coded Color Doppler Imaging (Power Doppler)

In the early 1990s, a new signal processing technique was introduced to internal medical investigations called power Doppler, UltrasoundAngio, color Doppler energy or amplitude-coded CDI (ACD; Rubin et al. 1993/1994; Preidler et al. 1995; Giovagnorio et al. 1995).

This technique encodes the amplitude of the power spectral density of the detected Doppler signal rather than the mean Doppler frequency shift (velocity and direction information) of conventional CDI ultrasound systems. Hence the information of an ACD image and its interpretation are fundamentally different: Neither velocity nor direction information of the blood flow is available, only the strength of the reflected Doppler signal amplitude is color coded, which is sometimes called the ‘energy’ of the Doppler signal.

However, this technique has opened new areas of clinical applications: Detecting slow blood flows or perfusion in renal blood vessels or kidneys represents only a few examples of many important clinical uses (Turetschek et al. 1998).

Modern CDI systems already have the potential for an upgrade to an ACD system if they contain an autocorrelation analyzer (Fig. 1.5).

The Doppler frequency shift signals are complex functions in a mathematical sense (the I-component corresponds to the real part, the Q-component to the imaginary part of the signal). This simplifies and shortens the signal processing in the complex autocorrelation module to get an average value for the Doppler frequency shift in real-time:

(4)

where ACFim is the imaginary part, ACFre the real part of the complex autocorrelation function.

This is the information displayed by conventional CDI systems.

In an ACD system, the complex autocorrelation function ACF (t) and the results of Eq. 4 are needed for certain time points t to calculate the power spectral density P(Δf) using the Wiener-Khintchine theorem:

(5)

This means that the Fourier transformation of the autocorrelation function is equal to the power spectral density function. Equation 5 is the mathematical description of this relationship, but in ACD systems, an integration from minus infinity to infinity cannot be performed in real-time. The solution to this problem is to replace the integration by a summation of the echo signals that are detected from all transmitted scan-lines. This can be done in real-time regardless of the large number of calculations necessary for creating only one image frame.

Figure 1.8 shows a plot of the power spectral density vs frequency. The integral over the slanted-fill curve (area) is the information that is finally displayed with ACD. The power spectral density measured at the ordinate is the squared amplitude of the signal. In ACD