Principles of Echocardiographic Image Acquisition and Doppler Analysis
An understanding of the basic principles of ultrasound imaging and Doppler echocardiography is essential both during data acquisition and for correct interpretation of the ultrasound information. Although, at times, current instruments provide instantaneous images so clear and detailed that it seems as if we can “see” the heart and blood flow directly, in actuality, we always are looking at images and flow data generated by complex analyses of ultrasound waves reflected and backscattered from the patient’s body. Knowledge of the strengths, and more importantly, the limitations, of this technique is critical for correct clinical diagnosis and patient management. On the one hand, echocardiography can be used for decision making with a high degree of accuracy in a variety of clinical settings. On the other hand, if an ultrasound artifact is mistaken for an anatomic abnormality, a patient might undergo needless, expensive, and potentially risky other diagnostic tests or therapeutic interventions.
In this chapter, a brief (and necessarily simplified) overview of the basic principles of cardiac ultrasound imaging and flow analysis is presented. The reader is referred to the Suggested Reading at the end of the chapter for more information on these subjects. Because the details of image processing, artifact formation, and Doppler physics become more meaningful with experience, some readers may choose to return to this chapter after reading other sections of this book and after participating in some echocardiographic examinations.
Sound waves are mechanical vibrations that induce alternate refraction and compression of any physical medium through which they pass (Fig. 1-1). Like other waves, sound waves are described in terms of (Table 1-1):
Figure 1–1 Schematic diagram of an ultrasound wave.
Frequency (f) is the number of ultrasound waves in a 1-second interval. The units of measurement are hertz, abbreviated Hz, which simply means cycles per second. A frequency of 1000 cycles/s is 1 kilohertz (KHz), and 1 million cycles/s is 1 megahertz (MHz). Humans can hear sound waves with frequencies between 20 Hz and 20 kHz; frequencies higher than this range are termed ultrasound. Diagnostic medical ultrasound typically uses transducers with a frequency between 1 and 20 MHz.
The speed that a sound wave moves through the body, called the velocity of propagation (c), is different for each type of tissue. For example, the velocity of propagation in bone is much faster (about 3000 m/s) than in lung tissue (about 700 m/s). However, the velocity of propagation in soft tissues, including myocardium, valves, blood vessels, and blood is relatively uniform, averaging about 1540 m/s.
Figure 1–2 Transducer frequency versus wavelength and penetration of the ultrasound signal in soft tissue.
Wavelength has been plotted inversely to show that resolution increases with increasing transducer frequency while penetration decreases. The specific wavelengths for transducer frequencies of 1, 2.5, 3.5, 5, and 7.5 MHz are shown.
The acoustic pressure, or amplitude, of an ultrasound wave indicates the energy of the ultrasound signal. Power is the amount of energy per unit time. Intensity (I) is the amount of power per unit area:
This relationship shows that if ultrasound power is doubled, intensity is quadruped. Instead of using direct measures of pressure energy, ultrasound amplitude is described relative to a reference value using the decibel scale. Decibels (dB) are familiar to all of us as the standard description of the loudness of a sound. Decibels are logarithmic units based on a ratio of the measured amplitude (A2) to a reference amplitude (A1) such that:
A simple rule to remember is that a 6-dB change represents a doubling or halving of the signal amplitude or that a 40-dB change represents a 100 times difference in amplitude (Fig. 1-3). If acoustic intensity is used instead of amplitude, the constant 10 replaces 20 in the equation so that a 3 dB changes represents doubling, and a 20 dB change indicates a 100-fold difference in amplitude. Either of these decibel scales may be used to refer to transmitted or received ultrasound waves or to describe attenuation effects. The advantages of the decibel scale are that a very large range can be compressed into a smaller number of values, and that low-amplitude (weak) signals can be displayed alongside very high-amplitude (strong) signals. In an echocardiographic image, amplitudes typically range from 1 to 120 dB. The decibel scale is the standard format both for echocardiographic image display and for the Doppler spectral display, although other amplitude scales may be an option.
Figure 1–3 Graph of the decibel scale.
The logarithmic relationship between the decibel scale (horizontal axis) and the amplitude ratio (vertical axis) is seen. A doubling or halving of the amplitude ratio corresponds to a 6-dB change, and a 100-fold difference in amplitude corresponds to a 20-dB change.
Propagation of ultrasound waves in the body to generate ultrasound images and Doppler data depends on a tissue property called acoustic impedance (Table 1-2). Acoustic impedance (Z) depends on tissue density (ρ) and on the propagation velocity in that tissue (c):
Although the velocity of propagation differs between tissues, tissue density is the primary determinant of acoustic impedance for diagnostic ultrasound. Lung tissue has a very low density compared to bone, which has a very high density. Soft tissues, such as blood and myocardium, have much smaller differences in tissue density and acoustic impedance. Acoustic impedance determines the transmission of ultrasound waves through a tissue; differences in acoustic impedance result in reflection of ultrasound waves at tissue boundaries.
Figure 1–4 Diagram of the interaction between ultrasound and body tissues.
Doppler analysis is based on the scattering of ultrasound in all directions from moving blood cells with a resulting change in frequency of the ultrasound received at the transducer. 2D imaging is based on reflection of ultrasound from tissue interfaces (specular reflectors). Attenuation limits the depth of ultrasound penetration. Refraction, a change in direction of the ultrasound wave, results in imaging artifacts.
The basis of ultrasound imaging is reflection of the transmitted ultrasound signal from internal structures. Ultrasound is reflected at tissue boundaries and interfaces, with the amount of ultrasound reflected dependent on the:
Smooth tissue boundaries with a lateral dimension greater than the wavelength of the ultrasound beam act as specular, or “mirrorlike,” reflectors. The amount of ultrasound reflected is constant for a given interface, although the amount received back at the transducer varies with angle because (like light reflected from a mirror) the angle of incidence and reflection is equal. Thus, optimal return of reflected ultrasound occurs at a perpendicular angle (90°). Remembering this fact is crucial for obtaining diagnostic ultrasound images. It also accounts for ultrasound “dropout” in a two-dimensional (2D) or three-dimensional (3D) image when too little or no reflected ultrasound reaches the transducer resulting from a parallel alignment between the ultrasound beam and tissue interface.
Scattering of the ultrasound signal, instead of reflection, occurs with small structures, such as red blood cells suspended in fluid, because the radius of the cell (about 4 µm) is smaller than the wavelength of the ultrasound signal. Unlike a reflected beam, scattered ultrasound energy may be radiated in all directions. Only a small amount of the scattered signal reaches the receiving transducer, and the amplitude of a scattered signal is 100 to 1000 times (40-60 dB) less than the amplitude of the returned signal from a specular reflector. Scattering of ultrasound from moving blood cells is the basis of Doppler echocardiography.
Although experimental studies show differences in backscattering with changes in hematocrit, variation over the clinical range has little effect on the Doppler signal. Similarly, the size of red blood cells and the compressibility of blood cells and plasma do not change significantly. Thus, the primary determinant of scattering is transducer frequency.
Scattering also occurs within tissues, such as the myocardium, from interference of backscattered signals from tissue interfaces smaller than the ultrasound wavelength. Tissue scattering results in a pattern of speckles; tissue motion can be measured by tracking these speckles from frame to frame, as discussed in Chapter 4.
Ultrasound waves can be refracted—deflected from a straight path—as they pass through a medium with a different acoustic impedance. Refraction of an ultrasound beam is analogous to refraction of light waves as they pass through a curved glass lens (e.g., prescription eyeglasses). Refraction allows enhanced image quality by using acoustic “lenses” to focus the ultrasound beam. However, refraction also occurs in unplanned ways during image formation, resulting in ultrasound artifacts, most notably the “double-image” artifact.
Attenuation is the loss of signal strength as ultrasound interacts with tissue. As ultrasound penetrates into the body, signal strength is progressively attenuated because of absorption of the ultrasound energy by conversion to heat, as well as by reflection and scattering. The degree of attenuation is related to several factors, including the:
The attenuation coefficient (α) for each tissue is related the decrease in ultrasound intensity (measured in dB) from one point (I1) to a second point (I2) separated by a distance (l) as described by the equation:
The attenuation coefficient for air is very high (about 1000×) compared to soft tissue so that any air between the transducer and heart results in substantial signal attenuation. This is avoided on transthoracic examinations by use of a water-soluble gel to form an airless contact between the transducer and the skin; on transesophageal echocardiography (TEE) examination, attenuation is avoided by maintaining close contact between the transducer and esophageal wall. The air-filled lungs are avoided by careful patient positioning and the use of acoustic “windows” that allow access of the ultrasound beam to the cardiac structures without intervening lung tissue. Other intrathoracic air (e.g., pneumomediastinum, residual air after cardiac surgery) also results in poor ultrasound tissue penetration because of attenuation, resulting in suboptimal image quality.
The power output of the transducer is directly related to the overall degree of attenuation. However, an increase in power output may cause thermal and mechanical bioeffects as discussed in Bioeffects and Safety, p. 27.
Overall attenuation is frequency-dependent such that lower ultrasound frequencies penetrate deeper into the body than higher frequencies. The depth of penetration for adequate imaging tends to be limited to approximately 200 wavelengths. This translates roughly into a penetration depth of 30 cm for a 1-MHz transducer, 6 cm for a 5-MHz transducer, and 1.5 cm for a 20-MHz transducer, although diagnostic images at depths greater than these postulated limits can be obtained with state-of-the-art equipment. Thus, attenuation, as much as resolution, dictates the need for a particular transducer frequency in a specific clinical setting. For example, visualization of distal structures from the apical approach in a large adult patient often requires a low-frequency transducer. From a TEE approach, the same structures can be imaged (at better resolution) with a higher-frequency transducer. The effects of attenuation are minimized on displayed images by using different gain settings at each depth, an instrument control called time-gain (or depth-gain) compensation.
Ultrasound transducers use a piezoelectric crystal both to generate and to receive ultrasound waves (Fig. 1-5). A piezoelectric crystal is a material (such as quartz or a titanate ceramic) with the property that an applied electric current results in alignment of polarized particles perpendicular to the face of the crystal with consequent expansion of crystal size. When an alternating electric current is applied, the crystal alternately compresses and expands, generating an ultrasound wave. The frequency that a transducer emits depends on the nature and thickness of the piezoelectric material.
Figure 1–5 Schematic diagram of an ultrasound transducer.
The piezoelectric crystal both produces and receives ultrasound signals, with the electric input-output transmitted to the instrument via the cable. Damping material allows a short pulse length (improved resolution). The shape of the piezoelectric crystal, an acoustic lens, or electronic focusing (with a phased-array transducer) are used to modify beam geometry. The material of the transducer surface provides impedance matching with the skin. The ultrasound pulse length for 2D imaging is short (1-6ms), typically consisting of two wavelengths (λ). “Ring down”—the decrease in frequency and amplitude in the pulse—depends on damping and determines bandwidth (the range of frequencies in the signal).
Conversely, when an ultrasound wave strikes the piezoelectric crystal, an electric current is generated. Thus, the crystal can serve both as a “receiver” and as a “transmitter.” Basically, the ultrasound transducer transmits a brief burst of ultrasound and then switches to the “receive mode” to await the reflected ultrasound signals from the intracardiac acoustic interfaces. This cycle is repeated temporally and spatially to generate ultrasound images. Image formation is based on the time delay between ultrasound transmission and return of the reflected signal. Deeper structures have a longer time of flight than shallower structures, with the exact depth calculated based on the speed of sound in blood and the time interval between the transmitted burst of ultrasound and return of the reflected signal.
The burst, or pulse, of ultrasound generated by the piezoelectric crystal is very brief, typically 1 to 6 µs, because a short pulse length results in improved axial (along the length of the beam) resolution. Damping material is used to control the ring-down time of the crystal and, hence, the pulse length. Pulse length also is determined by frequency because a shorter time is needed for the same number of cycles at higher frequencies. The number of ultrasound pulses per second is called the pulse repetition frequency, or PRF. The total time interval from pulse to pulse is called the cycle length, with the percent of the cycle length used for ultrasound transmission called the duty factor. Ultrasound imaging has a duty factor of about 1% compared to 5% for pulsed Doppler and 100% for continuous-wave (CW) Doppler. The duty factor is a key element in the patient’s total ultrasound exposure as discussed in Bioeffects and Safety, p. 27.
The range of frequencies contained in the pulse is described as its frequency bandwidth. A wider bandwidth allows better axial resolution because of the ability of the system to produce a narrow pulse. Transducer bandwidth also affects the range of frequencies that can be detected by the system with a wider bandwidth, which allows better resolution of structures distant from the transducer. The stated frequency of a transducer represents the center frequency of the pulse.
The simplest type of ultrasound transducer is based on a single piezoelectric crystal (Table 1-3). Alternate pulsed transmission and reception periods allow repeated sampling along a single line, with the sampling rate limited only by the time delay needed for return of the reflected ultrasound wave from the depth of interest. An example of using the transducer for simple transmission-reception along a single line is an A-mode (amplitude versus depth) or M-mode (depth versus time) cardiac recording when a high sampling rate is desirable.
Formation of more complex images uses an array of ultrasound crystals arranged to provide a 2D tomographic or 3D volumetric data set of signals. Each element in the transducer array can be controlled electronically both to direct the ultrasound beam across the region of interest and to focus the transmitted and received signals. Echocardiographic imaging uses a sector scanning format with the ultrasound signal originating from a single location (the narrow end of the sector), resulting in a fanlike shape of the image. Sector scanning is optimal for cardiac applications because it allows a fast frame rate to show cardiac motion and a small transducer size (aperture or “footprint”) to fit into the narrow acoustic windows used for echocardiography. Three-dimensional imaging transducers are discussed in Chapter 4.
Most transducers can provide simultaneous imaging and Doppler analysis, for example, 2D-imaging and a superimposed color Doppler display. Quantitative Doppler velocity data are recorded with the image “frozen” or with only intermittent image updates, with the ultrasound crystals used to optimize the Doppler signal. Although CW Doppler signals can be obtained using two elements of combined transducer, use of a dedicated nonimaging transducer with two separate crystals (with one crystal continuously transmitting and the other continuously receiving the ultrasound waves) is recommended when accurate high-velocity recordings are needed. The final configuration of a transducer depends on transducer frequency (higher-frequency transducers are smaller) and beam focusing, as well as the intended clinical use, for example, transthoracic versus TEE imaging.
An unfocused ultrasound beam is shaped like the light from a flashlight, with a tubular beam for a short distance that then diverges into a broad cone of light (Fig. 1-6). Even with current focused transducers, ultrasound beams have a 3D shape that affects measurement accuracy and contributes to imaging artifacts. Beam shape and size depend on several factors, including:
Figure 1–6 Schematic diagram of beam geometry for an unfocused (left) and focused (right) transducer.
The length of the near zone and the divergence angle in the far field depend on transducer frequency and aperture. The focal zone of a focused transducer can be adjusted, but beam width still depends on depth. Side lobes (and grating lobes with phased-array transducers) occur with both focused and unfocused transducers and, like the central beam, are 3D.
Aperture size and shape and beam focusing can be manipulated in the design of the transducer, but the effects of frequency and depth are inherent to ultrasound physics. For an unfocused beam, the initial segment of the beam is columnar in shape (near field Fn) with a length dependent on the diameter D of the transducer face and wavelength (λ):
For a 3.5-MHz transducer with a 5-mm diameter aperture, this corresponds to a columnar length of 1.4 cm. Beyond this region, the ultrasound beam diverges (far field), with the angle of divergence θ determined as:
This equation indicates a divergence angle of 6° beyond the near field, resulting in an ultrasound beam width of about 4.4 cm at a depth of 20 cm for this 3.5-MHz transducer. With a 10-mm diameter aperture, Fn would be 5.7 cm and beam width at 20 cm would be about 2.5 cm (Fig. 1-7).
Figure 1–7 Transducer frequency versus near zone length and divergence angle.
Transducer frequency is shown on the horizontal axis with the length of the near zone shown in yellow and divergence angle in blue for unfocused 5- (squares) and 10-mm (triangles) diameter aperture transducers. Equations (1-6) and (1-7) were used to generate these curves.
The shape and focal depth (narrowest point) of the primary beam can be altered by making the surface of the piezoelectric crystal concave or by the addition of an acoustic lens. This allows generation of a beam with optimal characteristics at the depth of most cardiac structures, but again, divergence of the beam beyond the focal zone occurs. Some transducers allow manipulation of the focal zone during the examination. Even with focusing, the ultrasound beam generated by each transducer has a lateral and an elevational dimension that depends on the transducer aperture, frequency, and focusing. Beam geometry for phased-array transducers also depends on the size, spacing, and arrangement of the piezoelectric crystals in the array.
In addition to the main ultrasound beam, dispersion of ultrasound energy laterally from a single-crystal transducer results in formation of side lobes at an angle θ from the central beam where sin θ = m λ/D, and m is an integer describing sequential side lobes (i.e., 1, 2, 3, and so on) (Fig. 1-8). Reflected or backscattered signals from these side lobes may be received by the transducer, resulting in image or flow artifacts. With phased-array transducers, additional accessory beams at an even greater angle from the primary beam, termed grating lobes, also occur as a result of constructive interference of ultrasound wave fronts. Both the side lobes and the grating lobes affect the lateral and elevational resolution of the transducer.
Figure 1–8 Transducer beam side lobes.
Top: This diagram shows that side lobes occur at the points where the distances traversed by the ultrasound pulse from each edge of the crystal face differ by exactly one wavelength. The distance from the left edge of the crystal (P1) to the position of side lobe 1 is exactly one wavelength (λ) longer than the distance from the extreme right edge of the crystal (P2) to the position of side lobe 1. Bottom: The beam intensity plot formed by sweeping along an arc at focal length F. (From Geiser EA: Echocardiography: physics and instrumentation. In Skorton DJ, Schelbert AR, Wolf GL, Brundage BH [eds]: Marcus Cardiac Imaging, 2nd ed. Philadelphia: WB Saunders, 1996, p 280. Used with permission.)
Figure 1–9 Axial, lateral, and elevational slice thickness in three dimensions for a phased-array transducer ultrasound beam.
A, Axial resolution along the direction of the beam is independent of depth; lateral resolution and elevational resolution are strongly depth dependent. Lateral resolution is determined by transmit and receive focus electronics; elevational resolution is determined by the height of the transducer elements. At the focal distance, axial is better than lateral and is better than elevational resolution. B, Elevational resolution profile with an acoustic lens across the transducer array produces a focal zone in the slice thickness direction. (From Bushberg JT, et al: The Essential Physics of Medical Imaging. Philadelphia: Lippincott Williams & Wilkins, 2002, Fig. 16-21).
Of these three, axial resolution is most precise, so quantitative measurements are made most reliably using data derived from a perpendicular alignment between the ultrasound beam and structure of interest. Axial resolution depends on the transducer frequency, bandwidth, and pulse length but is independent of depth (Table 1-4). Determination of the smallest resolvable distance between two specular reflectors with ultrasound is complex but is typically about twice the transmitted wavelength; higher-frequency (shorter-wavelength) transducers have greater axial resolution. For example, with a 3.5 MHz transducer, axial resolution is about 1 mm, versus 0.5 mm with a 7.5 MHz transducer. A wider bandwidth also improves resolution by allowing a shorter pulse, thus avoiding overlap between the reflected ultrasound signals from two adjacent reflectors.
Beam width (focusing) at each depth∗
Aperture (width) of transducer
Side and grating lobe levels
Beam width in elevational plane
Lateral resolution varies with the depth of the specular reflector from the transducer, primarily related to beam width at each depth. In the focal region where beam width is narrow, lateral resolution may approach axial resolution, and a point target will appear as a point on the 2D image. At greater depths, beam width diverges so a point target results in a reflected signal as wide as the width of the beam, which accounts for “blurring” of images in the far field. If the 2D image is examined carefully, progressive widening of the echo signals from similar targets along the length of the ultrasound beam can be appreciated (Fig. 1-10). Erroneous interpretations occur when the effects of beam width are not recognized. For example, beam width artifact from a strong specular reflector may appear to be an abnormal linear structure. Other factors that affect lateral resolution are transducer frequency, aperture, bandwidth, and side and grating lobe levels.
Figure 1–10 Beam width effect on 2D imaging.
2D echocardiographic view of the LV from an apical approach. The effect of beam width can be appreciated by comparing the length of reflections from point targets near and at greater distances from the transducer as shown by the arrows.