The pulse–echo experiment

To illustrate how ultrasound imaging works, the acoustic “pulse–echo” experiment can be used:

1. A short electric pulse is applied to a piezoelectric crystal. This electric field will induce a shape change of the crystal through reorientation of its polar molecules. Thus, the application of an electric field deforms the crystal.
   
2. The deformation of the piezoelectric crystal induces a local compression of the tissue with which the crystal is in contact. Thus, the superficial tissue layer is briefly compressed resulting in an increase in local pressure (acoustic pressure) (Figure 1.1).
   
3. Due to an interplay between tissue elasticity and inertia, this local tissue compression (with subsequent decompression or rarefaction) propagates away from the piezoelectric crystal at a speed of approximately 1530 m/s in human soft tissue (Figure 1.2). This is called the acoustic wave. The rate of compression/decompression determines the frequency of the wave and typically is between 1 and 15 MHz for human diagnostic ultrasonic imaging. As these frequencies cannot be perceived by the human ear, these waves are named “ultrasonic” (the range of human perceivable frequencies is between 20 and 20 kHz). The spatial distance between subsequent compressions is called the wavelength (λ) and relates to the frequency (f) and sound velocity (c) as: λ · f = c. During propagation, acoustic energy is lost mostly as a result of absorption resulting in a reduction in amplitude of the wave with propagation distance. The shorter the wavelength (i.e., the higher the frequency), the faster the particle motion and the larger the absorption effects. Higher frequency waves will thus attenuate more and penetrate less deeply into the tissue. This explains why the “high” frequency probes have less penetration.
   
4. Spatial changes in tissue density or tissue elasticity will result in a disturbance of the propagating compression (i.e., acoustic) wave and will cause part of the energy in the wave to be reflected. These so‐called “specular reflections” occur, for example, at the interface between different types of tissue (i.e., acoustic impedance difference), such as between blood and myocardium. The reflected waves behave similarly to optic waves as the direction is determined by the angle between the reflecting surface and the incident wave. This is similar to the reflection of optic waves on the water surface. When the spatial dimensions of the changes in density or compressibility become small relative to the wavelength (i.e., below ~100 μm), these inhomogeneities will cause part of the energy in the wave to be scattered and transmitted in different directions. Part of the scattered energy is transmitted back in the direction of source and is called backscatter. Both the specular and backscattered reflections propagate back towards the piezoelectric crystal.
   
5. When the reflected waves reach the piezoelectric crystal this causes deformation and results in relative motion of its (polar) molecules and generation of an electric field, which can be detected and measured. The amplitude of this electric signal is directly proportional to the amount of compression of the crystal, which is determined by the amplitude of the reflected/backscattered waves. This electric signal is the radiofrequency (RF) signal and can be represented as the amplitude of the reflected ultrasound wave as a function of time (Figure 1.3). Because reflections occurring further away from the transducer need to propagate further, they will be received later. As such, the time axis in Figure 1.3 can be replaced by the propagation distance of the wave (i.e., depth). The signal detected by the transducer is typically electronically amplified. The amount of amplification has a preset value but can be modified on an ultrasound system by using the “gain” button. Importantly, the overall gain will amplify both the signal and potential noise and will thus not impact the signal‐to‐noise ratio.

In the example shown in Figure 1.3, taken from a water tank experiment, two strong specular reflections can be observed (around 50 and 82 μs) while the lower amplitude reflections in between are scatter reflections. In clinical echocardiography, the most obvious specular reflection is the strong reflection coming from the pericardium observed in the parasternal views as a consequence of the high acoustic impedance difference. The direction of propagation of the specular reflection is determined by the angle between the incident wave and the reflecting surface. Thus, the strength of the observed reflection will depend on: (i) the exact transducer position; (ii) orientation with respect to the pericardium; and (iii) the acoustic impedance difference between the two structures. Indeed, for given transducer positions/orientations, the strong specular reflection might propagate in a direction not detectable by the transducer. For this reason, the pericardium typically does not show as bright in the images taken from an apical transducer position. In contrast, scatter reflections are not angle dependent and will always be visible for a given structure independent of the exact transducer position.

The total duration of the above described pulse–echo experiment is about 100 μs when imaging at 5 MHz The reflected signal in Figure 1.3 is referred to as an A‐mode image (“A” referring to “amplitude”) and is the most fundamental form of imaging given it tells us something about the acoustic characteristics of the materials in front of the transducer. For example, Figure 1.3 clearly shows that at distance of ~3.7 cm in front of the transducer the propagation medium changes density and/or compressibility, with a similar change occurring at a distance of ~6.3 cm (these distances correspond to 50 and 82 μs, respectively, times 1530 m/s – which is the total propagation distance of the wave – divided by 2 as the wave has to travel back and forth). The 2.6 cm of material in between these strong reflections is acoustically inhomogeneous (i.e., shows scatter reflections) and thus contains local (very small) fluctuations in mass density and/or compressibility, while the regions closer and further away from the transducer do not cause significant scatter and would thus be acoustically homogeneous. Indeed, this A‐mode image was taken from a 2.6‐cm thick tissue‐mimicking material (i.e., gelatin in which small graphite particles were dissolved) put in a water tank.

Grayscale encoding

Since the A‐mode image just presented is not visually attractive, the RF signal resulting from a pulse–echo experiment is further processed:

1. Envelope detection: The high‐frequency information of the RF signal is selected by detecting the envelope of the signal (Figure 1.4). This process is also referred to as “demodulation.”
   
2. Grayscale encoding: The signal is subdivided as a function of time in small intervals. Each pixel/voxel is attributed a number, defined by the local amplitude of the signal, ranging between 0 and 255 (2⁸ or 8‐bit image). “0” represents “black,” “255” represents “white,” and a value in between is represented by a grayscale. By definition, bright pixels correspond to high‐amplitude reflections (i.e., high acoustic impedance difference). This process is illustrated in Figure 1.4. Note that a different color encoding is also possible simply by coding different color intensities in a range of values between 0 and 255. On clinical scanners the color map can be easily selected and is a preference of the operator. Typically shades of blue or bronze can be used to represent the image. More modern ultrasound systems have higher resolution encoding with 12‐ or 16‐bit resolution images (i.e., encoding 4096 or 65,536 gray/color levels).
   
3. Attenuation correction: As wave amplitude decreases with propagation distance due to attenuation (mostly due to conversion of acoustic energy to heat, i.e., acoustic absorption), reflections from deeper structures are intrinsically smaller in amplitude and therefore show less bright. In order to give identical structures located at different distances from the transducer a similar gray value (i.e., reflected amplitude), compensation for this attenuation must occur. Thus, an attenuation profile as a function of distance from the transducer is assumed, which allows for automatic amplification of the signals from deeper regions – the so‐called automated time gain compensation (TGC), also referred to as depth gain compensation. As the preassumed attenuation profile might be incorrect mainly due to variable reflections and absorptions, sliders on the ultrasound scanner (TGC toggles) allow for manual correction of the automatic compensation and will result in more or less local amplification of the received signal as required to obtain a more homogenous brightness of the image. In this way, the operator can optimize local image brightness. It is recommended to start scanning using a neutral setting of these sliders, as attenuation characteristics will be patient and view specific. For every view the TGC can be optimized manually.
   
4. Log compression: In order to increase the image contrast in the darker (i.e., less bright) regions, gray values in the image may be redistributed according to a logarithmic curve (Figure 1.5). The characteristics of this compression (i.e., local contrast enhancement) can be changed on the ultrasound scanner (contrast or compression). These setting changes do not affect the ultrasound acquisition but only influence the visual representation of the created image. This is similar to contrast adaptation used in digital photography, where ambient lighting and/or contrast can be retrospectively enhanced independent of the acquisition settings. The setting on the system impacting the visual aspect of the image is the so‐called “dynamic range.” Adjustment of the dynamic range, or compression, settings changes the number of gray values used and therefore results in high contrast (i.e., almost black and white images without much gray) or low‐contrast images.

Image construction

In order to obtain an ultrasound image, the procedures of signal acquisition and post‐processing are repeated.

For conventional B‐mode imaging (“B” referring to “brightness”), the transducer can either be translated (Figure 1.6a) or tilted (Figure 1.6b) within a plane between two subsequent pulse–echo experiments. In this way, a conventional 2D cross‐sectional image is constructed. The same principle can be used for 3D imaging by moving the ultrasound beam in 3D space between subsequent acquisitions.

Alternatively, the ultrasound beam is transmitted in the same direction for each transmitted pulse. In that case, an image line is obtained as a function of time, which is particularly useful when looking at motion. This modality is therefore referred to as M‐mode (“M” referring to “motion”) imaging.

Side lobe artifacts

In the construction of an ultrasound image by focused waves, the assumption is made that all reflections originate from a region directly in front of the transducer. Although most of the ultrasound energy is indeed centered on an axis in front of the transducer, in practice part of the energy is also directed sideways (i.e., directed off‐axis). The former part of the ultrasound beam is called the main lobe whereas the latter is referred to as the side lobes (Figure 1.7, Videos 1.1 and 1.2).

Because the reflections originating from these side lobes are much smaller in amplitude than the ones coming from the main lobe, they can typically be neglected. However, image artifacts can arise when the main lobe is in an anechoic region (e.g., a cyst or inside the left ventricular cavity) causing the relative contribution of the side lobes to become significant. In this way, a small cyst or lesion may be more difficult to detect, as it appears brighter due to spillover of (side lobe) energy from neighboring regions. Similarly, when using contrast agents, the reflections resulting from small side lobes may become more significant as these agents strongly reflect ultrasound energy. As such, increased brightness may appear in regions adjacent to regions filled with contrast without contrast being present in the region.

Reverberation artifacts

When the reflected wave arrives at the transducer, part of the energy is converted to electric energy as described in the previous section. However, due to the acoustic impedance difference between the tissue and the transducer, another part of the wave is simply reflected on the transducer surface and will start propagating away from the transducer as if it was another ultrasound transmission. This secondary “transmission” will propagate in a way similar to that of the original pulse, which means that it is reflected by the tissue and detected again (Figure 1.8, Video 1.3).

These higher order reflections are called reverberations and give rise to ghost (mirror image) structures in the image (Video 1.4). These ghost images typically occur when strongly reflecting structures such as ribs or the pericardium are present in the image (Video 1.5). Similarly, as the reflected wave coming from the pericardium is very strong, its backscatter (i.e., propagating again towards the pericardium) will be sufficiently strong as well. This wave will reflect on the pericardium and can be detected by the transducer after the actual pericardial reflection arrives. In clinical practice, this causes a ghost image to be created behind the pericardial reflection that typically appears as a mirror image of the left ventricle around the pericardium in a parasternal long‐axis view.

Shadowing and dropout artifacts

When complete reflections occur due to high acoustic impedance difference, no acoustic energy is transmitted to more distal structures and – as a consequence – no reflections from these distal structures can be obtained. As a result, a very bright structure will appear in the image followed by a signal void, i.e., an acoustic shadow (Video 1.6). For example, when a metallic prosthetic valve has been implanted, the metal (being very dense and extremely stiff) can cause almost complete ultrasound reflections, resulting in an apparently anechoic region distal to the valve. This occurs because no ultrasound energy reaches the deeper regions. Similarly, some regions in the image may receive little ultrasound energy due to superficial structures blocking ultrasound penetration. Commonly, ribs (being more dense and stiffer than soft tissue) are strong reflectors at cardiac diagnostic frequencies and can impair proper visualization of some regions of the image. Or, even more obviously, the absence of ultrasound gel may allow air to be in contact with the transducer. As air has a very different acoustic impedance from biological tissues, the acoustic wave (at the emission frequency in medical imaging) cannot propagate at all. These artifacts are most commonly referred to as “dropout” and can only be avoided by changing the transducer position/orientation.

When signal dropout occurs at deeper regions only, the acoustic power transmitted can be increased. This will obviously result in more energy penetrating to deeper regions and will increase the overall signal‐to‐noise ratio of the image (in contrast to increasing the overall gain of the received signals as explained earlier). However, the maximal transmit power allowed is limited in order to avoid potential adverse biological effects. Indeed, at higher energy levels, ultrasound waves can cause tissue damage either due to cavitation (i.e., the formation of vapor cavities that subsequently implode and generate very high local pressures and temperatures) or tissue heating. The former risk is quantified in the mechanical index (MI) and should not pass a value of 1.9, while the latter is estimated through a thermal index (TI) and should not pass – depending on whether it is for nonobstetric or non‐neonatal applications – a value of a 1.0 for too long (the higher the value, the shorter the examination time should be). Overall, the acoustic power output of the system should not exceed 720 mW/cm², in accordance with US Food and Drug Administration (FDA) recommendations. This value is verified by the regulatory bodies to allow manufacturers to enter the market and thus does not have to be verified by the operator. Both MI and TI are always displayed on the monitor during scanning and will increase with increasing power output. The operator has to find a compromise between image quality (including penetration depth) and the risk of adverse biological effects. In case penetration is not appropriate at maximal transmit power, the operator should choose a transducer with a lower transmit frequency.

Phased‐array transducers

Rather than mechanically moving or tilting the transducers, as in the early‐generation ultrasound machines, modern ultrasound devices use electronic beam steering. To do this, an array of piezoelectric crystals is used. By introducing time delays between the excitation of different crystals in the array, the ultrasound wave can be sent in a specific direction without mechanical motion of the transducer (Figure 1.9). The RF signal for a transmission in a particular direction is then simply the sum of the signals received by the individual elements. These individual contributions can be filtered, scaled, and time delayed separately before summing. This process is referred to as beam forming and is a crucial element for obtaining high‐quality images. The scaling of the individual contributions is typically referred to as apodization and is critical in suppressing side lobes and thus avoiding the associated artifacts.

This concept can be generalized by creating a 2D matrix of elements that enables steering of the ultrasound beam in three dimensions. This type of transducer is referred to as a matrix array or 2D array transducer. Because each of the individual elements of such an array needs electrical wiring, manufacturing such a 2D array was technically challenging for many years because of the limitation of the thickness of the transducer cable. These obstacles have been overcome and 3D arrays are now available on high‐end ultrasound systems.

Second harmonic imaging

Wave propagation as illustrated in Figure 1.2 only happens when the amplitude of the ultrasound wave is relatively small (i.e., the acoustic pressures involved are small). Indeed, when the amplitude of the transmitted wave becomes significant, the shape of the ultrasound wave will change during propagation, as illustrated in Figure 1.10. The phenomenon of wave distortion during propagation is referred to as nonlinear wave propagation. This wave distortion results in the generation of harmonic frequencies which are integer multiples of the transmitted frequency. Transmitting a 1.7‐MHz ultrasound pulse will thus result in the spontaneous generation of frequency components of 3.4, 5.1, 6.8, and 8.5 MHz, and so on. These harmonic components become stronger with propagation distance. The rate at which the waveform distorts for a specific wave amplitude is tissue dependent and characterized by the nonlinearity parameter, β (or the so‐called “B/A” parameter).

The ultrasound scanner can be set up to receive only the second harmonic component through filtering of the received RF signal. If further post‐processing of the RF signal is done in exactly the same way as described earlier, a second harmonic image is obtained. Such an image typically has a better signal‐to‐noise ratio by avoiding clutter noise due to (rib or air) reverberation artifacts. As second harmonic waves are increasingly present with increasing depth, second harmonic imaging results in better images at deeper depths. This harmonic image is commonly used in patients with poor acoustic windows and poor penetration. Although harmonic imaging increases the signal‐to‐noise ratio, it has intrinsically poorer axial resolution as further discussed later in this chapter. Higher harmonics (i.e., third, fourth, etc.) are also present but typically fall outside the bandwidth of the transducer, which is the range of analyzable frequencies, and thus are not detected by the transducer. Harmonic imaging has become the default cardiac imaging mode for adult scanning on many systems. It is typically unnecessary to use harmonic imaging in young infants as it reduces axial resolution. Switching between conventional and harmonic imaging is done by changing the transmit frequency of the system. For lower frequency transmits, the default setting is usually a harmonic imaging mode, which is indicated on the display by showing both transmit and receive frequencies (e.g., 1.7/3.4 MHz). When a single frequency is displayed, the scanner is in a conventional (i.e., fundamental) imaging mode. For pediatric scanning, especially in smaller infants, fundamental imaging is the preferred mode due to its better spatial resolution, and presets for higher frequency probes are typically programmed in the fundamental frequency.

Contrast imaging

As blood is a poor reflector of ultrasound energy (i.e., relative homogeneity of the acoustic impedance), it shows dark in the image. For some applications, like myocardial perfusion assessment, it can be useful to artificially increase blood reflectivity. This can be achieved by using an ultrasound contrast agent. As air is a very strong reflector of ultrasound energy due its relative high compressibility and low density when compared with soft tissue, it often is used as a contrast agent. The injection of small air bubbles with diameters similar to those of red blood cells, significantly increases blood reflectivity. Agitated saline can be used or ultrasound contrast agents that contain encapsulated air bubbles to limit the diffusion of air in blood. Contrast imaging can be helpful for visualizing the endocardial border as it enhances the difference in gray value between the myocardium and the blood pool. It can be used in patients with poor image penetration to better visualize the endocardial border. Contrast injection can also be used for detecting shunts. As in pediatrics, right‐to‐left shunting can be present; typically carbon dioxide is the preferred gas to create agitated saline. Contrast agents can also be used to increase the brightness of the perfused myocardial tissue, although artifacts can be present which can make the interpretation of the perfusion images less obvious. At present, different contrast agents are commercially available for clinical use but regulatory approval for pediatric use differs between countries. In the presence of right‐to‐left shunting, contrast agents should be used cautiously and only by experienced operators.