Buzzwords in Sonography
Frederick W. Kremkau, Ph.D.
Professor and Director for Medical Ultrasound
Center for Applied Learning and School of Medicine
Wake Forest University
In diagnostic medical sonography, short pulses of ultrasound are sent into the body by a device called a transducer (Fig 1).
Each pulse travels down a straight path, interacting with the biological structures and producing a stream of echoes that is received and displayed as a straight line of dots on the sonographic display. This line of dots is called a scan line. Approximately 100 to 250 scan lines combine to yield a gray-scale presentation of anatomy on the display of a sonographic instrument (Fig. 2).
An image like that in Figure 3 can be generated in a fraction of a second. Thus, many images (called frames) can be generated per second, yielding real-time sonography capable of following motion as it occurs. This motion can be the motion of the tissue (e.g. pulsating vessels walls) or the motion of the moving scan plane as the sonographer manually slides the transducer over the surface of the patient.
Sonographic displays are of two types. The rectangular image in Fig. 3 is good for superficial imaging with its wide field of view up close. The pie-slice-shaped, sector image is good for deep imaging, providing a wide field of view deep while requiring a small contact area on the surface (Fig. 4). Several 2D images can be acquired and combined into a 3D volume (Fig. 5). This stored volume can then be presented in multiple ways. Figure 6 gives an example of surface rendering.
Sonographic instruments are organized into four primary parts (Fig. 7).
The transducer generates the ultrasound pulses and receives the returning echo streams that form the scan lines of each image. The transducer (T) is connected to the beam former. The beam former generates the electric voltage pulses that drive the transducer. The returning echo voltages from the transducer enter the beam former, are amplified and converted to digital form for processing in the signal processor. The signal processor converts the echo voltages to a simpler (amplitude) form and sends them on to the image processor where the scan lines form the image and are stored in image memory. Then the image is sent from memory to the flat-panel display for viewing. The functions of the beam former, signal processor and image processor are listed in Tables 1, 2 and 3. Details are found in Reference 1.
- Generate voltages that drive the transducer
- Determine PRF, coding, frequency, and intensity
- Scan, focus, and apodize the transmitted beam
- Amplify the returning echo voltages
- Compensate for attenuation
- Digitize the echo voltage stream
- Direct, focus, and apodize the reception beam
Table 1. Functions of the beam former.
- Bandpass Filtering
- Amplitude Detection (RF to video)
- Compression (dynamic range reduction)
Table 2. Functions of the signal processor.
- Scan conversion
- Panoramic imaging
- Spatial compounding
- 3D processing
- Storing Image Frames
- Cine loop
- Gray Scale
- Color Scale
- Digital to analog conversion
Table 3. Functions of the image processor.
Discussed now are several relatively recent and advanced functions (features) of contemporary sonographic technology as listed Table 4.
- Coded Excitation
- Harmonic Imaging
- Panoramic Imaging
- Spatial Compounding
- Volume Imaging
Table 4. Features of contemporary sonographic technology.
In conventional pulse-echo sonography, the beam former drives the transducer with a single-cycle voltage of the appropriate frequency (that desired for the generated ultrasound pulse). With coded excitation, a more complicated driving regime is used. This approach enables functions such as multiple transmission foci, separation of harmonic echo bandwidth from the transmitted pulse bandwidth, increased penetration, reduction of speckle with improved contrast resolution, and gray-scale imaging of blood flow (called B-flow). In straightforward pulsing, the pulser drives the transducer through the pulse delays with one voltage pulse per scan line. In coded excitation, ensembles of pulses drive the transducer to generate a single scan line. For example, instead of a single pulse, a series (such as three pulses, followed by a missing pulse [a gap], followed by two pulses, followed by another gap, followed by two pulses) could be used. Other examples are shown in Figure 8. A decoder in the receiving portion of the beam former recognizes and disassembles the coded sequence in the returning echoes and stacks up the individual pulses in the sequence to make a short, high-intensity echo out of them. The result is equivalent to having a much higher-intensity driving pulse or a much more sensitive receiving system. Thus for example in the case of blood flow, weak echoes from blood are imaged and flow can be seen in gray scale along with the much stronger tissue echoes (Figure 9).
Figure 8. Examples of coded pulse sequences. Each pulse consists of a cycle of pressure variation.
Coded excitation has been applied in radar for decades. A coded pulse is one that has internal amplitude, frequency, or phase modulation used for pulse compression. Pulse compression is the conversion, using a matched filter, of a relatively long coded pulse to one of short time duration, excellent resolution, and equivalent high intensity and sensitivity. A matched filter maximizes the signal-to-noise ratio of the returning signal. The longer the coded pulse, the higher the signal-to-noise ratio in matched-filter implementations will be. The intra-pulse coding is chosen to attain adequate axial resolution, and the pulse duration is chosen to achieve the desired sensitivity. The matched-filter decoding process can be thought of as a sliding correlation of the parts of the coded pulse with the matched filter. The result of this process is, in effect, a shorter and stronger pulse (Figure 10) yielding good resolution and sensitivity while conforming to transmitted pulse amplitude and intensity limitations imposed by technologic and safety considerations. Such coding schemes are called Barker codes. An even better match can be achieved by Golay codes that use pairs of transmitted pulses with the second being a bipolar sequence in which the latter portion of the pulse is the inverse of the first (Figure 10, J).
Figure 10. Matched filter decoding of coded pulse.
The dependence of propagation speed on pressure causes strong sound (pressure) waves to change shape as they travel (Figure 11). The reason for this is that the higher-pressure portions of the wave travel faster than the lower-pressure portions. This causes a wave that originally is shaped in a smooth curve form (called sinusoidal; illustrated in Figure 11, A) to progress toward a non-sinusoidal shape (Figure 11, C). Propagation in which speed depends on pressure and the wave shape changes is called
Figure 11. In nonlinear propagation, propagation speed depends on pressure.
nonlinear propagation. A continuous (not pulsed) sinusoidal waveform is characterized by a single frequency (equal to the number of cycles per second). Any other wave shape contains additional frequencies that are even and odd multiples of the original frequency. The original frequency is called the fundamental frequency. The even and odd multiples are called even and odd harmonics, respectively. A frequency analysis of the wave in Figure 11, A, would yield a single (fundamental) frequency such as 2 MHz. Parts B and C would reveal, in addition to the fundamental frequency, harmonics such as 4, 6, and 8 MHz. As the shape becomes less sinusoidal, the harmonics become stronger. Therefore they are stronger in part C than in part B. Using harmonic frequency echoes improves the quality of sonographic images.
A second type of filtering occurs with harmonic imaging, in which the fundamental (transmitted) frequency is filtered out and the second harmonic frequency echoes are passed. At this point the bandpass filter is centered at the second harmonic frequency with an appropriate bandwidth to include the bandwidth of the second harmonic echo signal (Figure 12, A to D). Harmonic imaging improves the image quality in three primary ways (Figure 12, E and F):
- The primary beam is much narrower, improving lateral resolution, because harmonics are generated only in the highest-intensity portion of the beam.
- Grating lobe artifacts are eliminated because these extra beams are not sufficiently strong to generate the harmonics.
- Because the harmonic beam is generated at a depth beyond where some of the artifactual problems occur (e.g., superficial reverberation), the image degradation that they cause is reduced or eliminated.
Because the fundamental and second harmonic bandwidths must fit within the overall transducer bandwidth (Figure 13, A), they must be reasonably narrow. This means that the corresponding ultrasound pulses must be somewhat longer than otherwise, causing some degradation in axial resolution. A solution to this degradation in image quality is to use pulse inversion, a technique that uses two pulses per scan line rather than one. The second pulse is the inverse of the first. The echo sequences from the two pulses (Figure 13, B) are added together to yield the resulting scan line. Fundamental frequency echoes cancel (Figure 13, C), and the second harmonic echoes remain (Figure 13, D). This technique allows broad-bandwidth, short pulses to be used so that detail resolution is not degraded (Figure 13, E). Instead, frame rate is reduced, with some degradation of temporal resolution.
Panoramic imaging provides a way to produce an image that has a wider field of view than what is available on an individual frame from a transducer. Panoramic imaging is achieved by manually sliding the transducer in a direction parallel to the scan plane, thus extending the scan plane. At the same time, the old echo information from previous frames is retained while the new echoes are added to the image in the direction in which the scan plane is moving. The result is a larger field of view allowing presentation of large organs and regions of anatomy on one image (Figure 14). The addition of the new echoes to the existing image as the transducer is moved requires their proper location relative to the existing image. This is accomplished by correlating locations of echoes common to adjacent frames (i.e., the overlap) so that the new information on the new frame is located properly (Figure 14, F and G).
Spatial compounding is a technique in which scan lines are directed in multiple directions by phasing so that structures are interrogated more than once by the ultrasound beam (Figure 15). Averaging sequential frames spatially, up to nine typically, improves the quality of the image in several ways:
- As in persistence (which is temporal averaging), speckle is reduced.
- Clutter caused by artifacts is reduced.
- Smooth (specular) surfaces are presented more completely because they are interrogated at more than one angle, increasing the probability that close to 90-degree incidence is achieved (which is necessary to receive echoes from them).
- Structures previously hidden beneath highly attenuating objects can be visualized.
For several years, an ultrasound imaging mode termed elastography has been studied. It has now become available commercially. By subjecting tissues to a small push (using a push on the transducer by the sonographer or using a high-amplitude ultrasound pulse) and then tracking the movement of the tissues, it is possible to estimate and depict tissue stiffness (since soft tissues will move more than hard ones). Essentially, elastography is the imaging version of palpation. It is commonly shown as a color overlay on top of the gray-scale image (Figure 16) and has been used clinically for cancer detection and characterization of small parts (breast, thyroid and prostate), for assessing the viability of the myocardium and for monitoring therapies that alter tissue composition such as ablation procedures.
Volume imaging is accomplished by acquiring many parallel two-dimensional (2D - slice imaging) scans (Figure 17, A) and then processing this 3D volume of echo information in appropriate ways for presentation on 2D displays. The multiple 2D frames are obtained by (1) manual scanning of the transducer, with position-sensing devices keeping track of scan-plane location and orientation, (2) automated mechanically scanned transducers, or (3) electronic scanning with 2D element-array transducers. Common ways of presenting the 3D echo data include surface renderings (Figure 17, B), 2D slices through the 3D volume, and transparent views. The advantage of the 2D slice presentation is that image-plane orientations can be presented that are impossible to obtain with conventional 2D scanning. Serial slice presentations like those in other imaging modes (MR and CT) can be presented (Figure 17, C-H). Surface renderings are popular in obstetric imaging.
Figure 17. Three-dimensional sonographic images.
Three-dimensional images can be acquired at rates of up to 30 volumes per second and thus are considered "live" or real-time presentations. The current tendency is to call this "4D imaging" where the fourth dimension is time. Although this term sounds fashionable for marketing purposes, it is inconsistent with previous terminology; that is, real-time 2D imaging is not called 3D. Nevertheless, the moniker has, to some extent, caught on.
More examples of presentation modes for volume imaging are shown in Figure 18
Figure 18. Various post-processing choices for presenting three-dimensional images.
Finally, miniaturization is making sonography more accessible for "point-of-care" application and further distribution throughout medical practice for diagnostic imaging (Figures 19 and 20).
Figures are from Reference 1 by permission.Reference 1. Kremkau, FW: Sonography Principles and Instruments, 8th Edition, Elsevier/Saunders, 2011.