Transducer Characteristics for Imaging
This section discusses the criteria for linking which properties of ultrasound imaging transducers and their formats need to be identified for various clinical applications.3 What follows is primarily applicable to clinically used imaging transducers that operate in the frequency range of 1 to 20 MHz. Transducers operating above this frequency are used for special applications such as intravascular imaging (see Figure 4, F and G) or preclinical imaging of small animals but are also included in the discussion wherever possible.
How well is the type of transducer suited to the “acoustic window” or location where it makes contact with the body to visualize the organs or tissues of interest? Standard acoustic windows provide an unobstructed view of an organ or region; many, by convention, have specific names, such as “transabdominal” or “parasternal long-axis,” so that images can be compared and described consistently.3 Typical windows are located in or on the following general regions of the body: head, chest, abdomen, pelvis, limbs, vessels, and various orifices of the body. Transducers can be associated with certain regions through Latin prefixes: “trans,” through or across; “intra,” into or inside; and “endo,” within, etc. An example is transthoracic, a category that includes transducers that image through the chest. Transcranial describes a transducer that images the head through the skull.
As already mentioned, for the transthoracic window, the phased array would be the most appropriate if the imaging task requires the transducer to be placed between the ribs; it is designed to fit into intercostal spaces and maximize the scanned area (format 2 of Figure 3). For most contact surfaces that are relatively flat and/or slightly deformable (eg, ones used for small parts or vascular imaging), the most general and frequently used transducer type is the linear array designed to make contact with flat surfaces, with the footprint decreasing in size with increasing frequency. Here rectangular and trapezoidal formats (1 and 4 of Figure 3) provide appropriate viewing areas. With abdominal imaging, to increase the viewing area with minimal increases of the contact area, convex arrays (Figure 4C) produce format 3 (shown in Figure 3) and are designed to make surface contact in deformable soft areas of the body.
Specialized transducers are designed to operate inside the body. These include transesophageal probes that are phased arrays suitable for manual manipulation within the esophagus, format 2 and transducer type B in Figure 4. A number of other specialty probes have also been developed for interventional or surgical use such as laparoscopic arrays and intracardiac arrays. These probes can be either linear or phased arrays, depending on the application and access windows. Several endo probes: such as endovaginal, endorectal, and endocavity (type D shapes), are functionally like end-fire phased arrays (format 2 and Figure 4B) or convex arrays (format 3 and Figure 4C) at the end of a small-diameter cylindrically shaped handle to fit in orifices and yet maximize the FOVs. Another example is the intravascular ultrasound transducer (Figure 4H), which is inserted in veins to produce an image plane, format 8 or scanned volume, format 9.
Resolution and Penetration
The selected scan depth allows viewing over the range of interest. Factors involved in imaging capability include the size of the active aperture (occult to the user, but typically a low F number [F#; focal depth/active aperture width] value of 1–2 is used), the transmit focal depth, and time-gain control settings available. Penetration is the minimum scan depth at which electronic noise is visible, despite optimization of available controls (usually at the deepest transmit focal setting and maximum gain), and electronic noise stays at a fixed depth even when the array is moved laterally. Penetration is primarily determined by the center frequency of the transducer: the higher the frequency, the shallower the penetration because the absorption of the ultrasound wave traveling through tissue increases with frequency.
A useful first approximation for estimating a depth of penetration (dp) for a given frequency is dp = 60/f cm-MHz, where f is given in megahertz. Thus, one might expect a 6-cm penetration from a 10-MHz center frequency transducer. As noted earlier, the absorption coefficient (acoustic power loss per unit depth) is a function of frequency and varies from tissue to tissue (values for soft tissues range from 0.6 to 1.0 dB/cm-MHz4). A more general term describing acoustic loss is the attenuation coefficient, which includes additional losses due to scattering and diffusion and hence is always greater than the absorption coefficient. The attenuation coefficient is highly patient and acoustic path dependent.
To optimize image resolution, users and manufacturers have worked on increasing the imaging frequencies for the various examination types. For example, some 30 years ago, people might have imaged the abdomen with a frequency of 2.25 MHz, whereas today the number is more often 3.5 MHz with some obstetric and gynecologic imaging reaching up to 5 MHz.5 Similarly, the last decade has seen a steady increase in breast imaging reaching the low teens in megahertz.
Transducer Properties and Imaging
Other criteria to be included in the above-discussed selection process are the transducer efficiency,2 transducer-system design, system signal-to-noise ratio, and, as already noted, absorption of the tissues being imaged. A major factor is absorption, the compositions and relative positions of different tissue types in the acoustic path. For example, a thick layer of adipose tissue will reduce penetration due to refractive or aberration errors in the acoustic path to the site of interest. Similarly increased amounts of amniotic fluid with fetal imaging enhance penetration and may permit the use of frequencies higher than those ordinarily used at the given scanning site.
The frequency range, or bandwidth,1,2,6 of the transducer will determine whether it can support B-mode imaging at different center frequencies and also operate in Doppler, harmonic, and color flow modes. With Doppler-based imaging modes, we often need to operate with lower frequencies than the B-mode frequency to minimize aliasing. With harmonic imaging, by definition, one uses a receive frequency that is a multiple (usually 2) of the transmitted frequency; hence the need for the wide bandwidth. The bandwidth and focusing properties will also influence image resolution. In clinical practice, it is essential to ensure that the image obtained can discern the smallest possible dimensions in both the lateral and axial directions.
Finally, the number of individual transducer elements is of interest because the number of active elements (with the exception of phased arrays or angularly scanned 2D arrays) determines the lateral extent or width of the image. For phased arrays, an increasing number of elements is associated with improved resolution and penetration depth. For 2D arrays (usually symmetric), the number of elements along the x and y directions determines the extent of the volume for linearly scanned arrays. For a 2D phased array, resolution and penetration increase with a greater number of elements along the x and y directions, but the angular shape or FOV remains the same, independent of the number of active elements used. The focusing in a fixed direction can indirectly affect imaging because focusing is positioned at only one depth and is poorer elsewhere. For 3D imaging, mechanically scanned 2D arrays suffer from the same fixed elevation focal depth limitation encountered in 2D imaging. In contrast, all the elements of fully populated 3D imaging or matrix arrays focus electronically at one point in both azimuth and elevation planes to provide far better resolution.
At the deepest depths, it is the maximum number of available active channels in the system that determines the resolution (along with strength of focusing and system noise). Spatial resolution is generally poorer (typically by a factor of 2) than the temporal resolution along scan lines; in the discussion presented here, resolution refers to spatial resolution, unless noted otherwise. For phased arrays, the number of channels usually corresponds to the maximum number of elements. As a general rule of thumb, because elements are typically on half-wavelength spacing, the more elements, the better the spatial resolution, which is inversely proportional to the active aperture in wavelengths. For example, a 64-element array, 32-wavelength aperture will have maximum spatial resolution 2-fold lower (wider beam) than that of a 128-element, 64-wavelength array. In the case of a linear array, which may have several hundred elements, the number of elements determines the lateral extent of the image, but it is the number of active channels that governs the resolution. For these 1D arrays, the resolution out of the imaging plane (also known as slice thickness) is poor except near the fixed elevation focal length. For 2D arrays, the spatial resolution is inversely proportional to the active apertures that form the sides of the 2D array. Two-dimensional arrays have superior resolution compared to 1D array focusing with fixed elevation focusing because true-point focusing can be achieved simultaneously in azimuth and elevation for 3D imaging.
Another way of looking at resolution is F#. The smaller the F#, the better the resolution.1 A simple estimate of the full (beam) width in millimeters, a common measure of resolution, neglecting absorption, is approximately F# × λ, where λ is wavelength (1.5 mm/μs/f [MHz]). For example, resolution would be 0.3 mm at 5 MHz for an F# = 1. Focal depths are also active aperture dependent. For example, for a 128-element 64-wavelength array, the deepest focal depth achieved at maximum aperture and an F# = 1 is F = F# × L = 64 wavelengths. The actual penetration or useable scan depth would, of course, be deeper than the maximum focal depth.
Matching Transducers to Clinical Applications
Now that transducer types and properties have been related to imaging and acoustic windows, they can contribute to the selection of transducers for specific clinical applications.3 The appropriateness of certain transducers to specific applications evolved historically and through specific tailored designs. The primary considerations are the target region of interest and its extent and the available acoustic windows needed for access.
When transducer arrays were initially introduced commercially for abdominal imaging (including obstetrics and gynecology) in the 1970s, they were of the linear array type (type A in Figure 4 with image format 1 of Figure 3). In most cases, the contact area with the patient was not a critical issue, and some of these linear arrays were quite long (eg, 8 cm) to cover, say, the third-trimester fetal head.3,5 However, it was soon realized that one could achieve sufficiently large coverage by the use of curved or convex arrays (type C in Figure 4) without incurring the penalty of having to manipulate the rather unwieldy linear array transducers.
The convex arrays (Figure 4C) are the tools of choice for most general 2D imaging applications involving the abdomen. The general form factor, related to ergonomic factors and the suitability of the transducer shape and FOV to the application, for abdominal 3D imaging is still evolving. Three key descriptors for these arrays are the footprint (overall aperture size), FOV, and radius of curvature (Figure 1C). The footprint describes the contact area usually in the form of a rectangle, circle or ellipse. Even though for abdominal imaging, access is not usually a concern, when these types of transducers are considered for new applications, window access is of primary importance. The radius of curvature and FOV (expressed in degrees of maximum angular coverage) are related to image extent and coverage. Advanced signal processing has been added to some systems to increase penetration; however, this feature is usually only available on certain probes.
For the mechanical 3D probes, the currently preferred form factor is a mechanically swept convex array (Figure 4G and format 6 in Figure 3); however, fully electronic convex 2D arrays are now becoming available. For these cases, two FOVs are given for the orthogonal scan directions. Alternatively, phased arrays, because of their small footprint and wide sector image format, are also used for abdominal imaging. Finally, 2D or matrix arrays are becoming increasingly prevalent for these applications because of their superior image quality, resolution, and ease of use.
The primary applications of this imaging grouping are cardiac scanning and examination of the liver from between the ribs. Simply because of the restrictive anatomy and the limited acoustic windows caused by the ribs and the often encroaching lungs, the transducer choice here is limited to phased arrays (Figure 4B). Even in this area, initial attempts were made to use linear arrays; however, these were rapidly dropped due to the rib shadowing and the superiority of phased array transducer format 2. For cardiac applications, the probes tend to have array dimensions on the order of 20 × 14 mm depending on the manufacturer. The patient contact area will be slightly larger. These numbers have evolved over the last 40 years and depend on a number of things, such as the general size of the patient population. Age is another consideration; rib spacing and the depth penetration needed vary as children mature into adults.
For noncardiac intercostal applications, the dimensions of the arrays are somewhat larger. As noted earlier, the existence of these anatomic limitations creates an upper performance limit for spatial resolution since resolution performance is inversely related to the size of the aperture, as explained above. In both cardiac and general intercostal imaging applications, the imaging depth is deep (depending on the patient size, it may be as deep as 24 cm), forcing the use of lower (1–3.5 MHz) frequencies and resulting in some further loss of imaging performance.
There is an interesting aspect of cardiac imaging that has had a profound effect on the nature of the probes. Due to the presence of the ribs and other acoustically hostile tissue in the ray path, echocardiography suffers from imaging artifacts due to reverberant noise. The introduction of harmonic imaging has proven to be highly successful in reducing this noise. As a consequence, the importance of transducer bandwidth has become critical in cardiac transducer design. Today, most cardiac systems transmit at frequencies between 1.5 and 2.0 MHz and, of course, receive signals at frequencies twice that range.
A major development in cardiac imaging was the implementation of fully populated 2D or matrix arrays (type E) containing thousands (typically 50 × 50 or so) of elements. These make possible real-time (4D) depiction of pyramidal volumes (format 7, Figure 3), visualization of arbitrary cut planes, and 4D cardiac imaging and color flow imaging. In addition, true electronic focusing in both the xz and yz planes provides superior resolution in comparison with all other 1D array transducers.
Superficial and Breast Imaging
This category refers to “superficial” imaging of carotids, leg veins, breasts, thyroids, testicles, etc and includes the categories of small parts, musculoskeletal, and peripheral vascular imaging. It is the last bastion of the application of linear arrays (type A), which formed the starting design type for the applications discussed earlier. In this clinical category, access is usually not an issue, and the sizes of the probes themselves can be small (because of the use of high 7- to 15-MHz frequencies and the resultant small element sizes). Musculoskeletal applications for imaging muscles, ligaments, and tendons also use arrays of this type. In the last 10 years, breast imaging has gone to very high frequencies (eg, 14 MHz), while imaging of the peripheral vasculature has remained at lower (about 3–11 MHz) values due to the need to include deeper leg veins and successful Doppler performance. Usually the capability of the array to add trapezoidal imaging (format 4) is a considerable advantage. As in abdominal imaging, 3D imaging with mechanically swept probes or electronic 2D arrays is now available for superficial and breast applications, greatly improving the coverage available and image quality. For applications involving imaging vasculature, some probes have advantages of including modes that enhance flow visualization.
Obstetrics and Gynecology
At the present time, mechanically scanned convex or linear arrays (types G and F) are used widely to provide 3D and 4D imaging of fetuses in vivo (formats 5–7). Matrix or fully populated 2D arrays (type E) are also available for this application (typically format 7).
For gynecology, specialized endo-array probe shapes are used (type D). Typically, the arrays are at the end of the probe (end-fire arrays) and are convex or curved arrays with wide FOVs (format 3); however, phased arrays in an endo-array package (type D) can also be used (format 2). Frequencies used are typically 5 MHz and higher. As in other applications, 2D arrays have been designed for 3D imaging of these cases.
Neonatal and Pediatric
Pediatric transducers tend to have smaller footprints than transducers used for adults, applications A–C, and operate at higher-frequency varieties (≥7 MHz) of those that are used for adults. Depending on the body region, types of transducers similar to those for adults are applicable. Phased arrays (type B) and 3D transducers (types E and G) are appropriate for cardiac imaging. Other arrays that are also useful for these clinical needs include static (2D) and, for 3D, mechanically swept linear arrays and convex arrays.
Intracavity probes constitute a large group of specialty transducers that are designed to image inside the body cavity. Transesophageal transducers are used to enable imaging of internal organs, especially the heart, from inside the esophagus (see Figure 5). They use higher frequencies (≥5 MHz) and are implemented as phased arrays with manipulators and motors to adjust the orientation of the transducer. Miniature transesophageal 2D arrays offer electronic scanning for 3D and 4D imaging.
Transducers can be highly specialized for viewing usually within body openings or vessels. Intracardiac phased arrays are inserted through a vessel to gain access to the inner chambers of the heart. Surgical specialty probes include laparoscopic arrays inserted through small incisions to image and aid in laparoscopic surgery (similar to endo probes); these are remarkable for their FOV despite small diameters. Intraoperative arrays are specially shaped to be placed on vessels, organs, and regions made accessible during open surgery (see Figure 5). Others in this class are surgical and interventional probes with unique shapes (see Figure 5).
As already noted, the probes that are inserted into the body are designed to fit through small openings and have a wide FOV (90°–150°). These probes include transrectal (or endorectal) for imaging of the pelvic region using the anus for access and the already described endovaginal (also called transvaginal) for imaging the female pelvis and reproductive organs using the vagina as entry for gynecologic and obstetric applications. These endo probes, described earlier, are cylindrical to fit into small orifices and have convex arrays (typically 3–9 MHz) at their ends with large fields of view, biplanes, or mechanically swept convex arrays. Probes for urologic applications include the biplane type.
A unique transducer is the biplane probe, which consists of two orthogonal arrays producing images in planes xz and yz. Typically the arrays are small (8–12 mm) and of the convex type. Each format and transducer would correspond to those of a single–imaging plane transducer such as format 3 of Figure 3 and the convex array of Figure 4C. However, sector or linear array formats are also possible, depending on the transducer construction, so that several combinations can be used in practice. Alternatively, a subset of the imaging capability of a 2D array is the simultaneous presentation of two orthogonal 2D images.
Intravascular transducers are inserted into blood vessels to image the vessel walls for various pathologic conditions (type H and formats 8 and 9). They are most often mechanically rotated single transducers with frequencies greater than 20 MHz and dedicated imaging systems, although there are also tiny (about 2-mm-diameter) arrays designed for this purpose.
Transcranial imaging of the brain and its vasculature is conducted through limited acoustic windows through the skull such as the temples or eyes. Transorbital arrays are high-frequency (typically >20 MHz) ophthalmologic transducers and are used to image the eye or use the eye as an acoustic window. Transcranial probes are usually lower-frequency (1–4 MHz) phased arrays used to image blood vessels within the skull through the temples as windows.
Many ultrasound imaging transducers are designed to operate in certain regions of the body for specific applications. A primary objective of this article is to provide a systematic approach that would aid in matching a transducer to a clinical application, starting with the acoustic window and the region and depth to be imaged. To this end, a checklist for selecting a transducer is given in Table 1.
As indicated earlier, the first consideration for imaging a target region or organs is the access: the intended acoustic window. The transducer type must provide access through the selected acoustic window. The transducer type is linked with the image format, and common selections previously discussed include the linear, phased, convex, and 2D arrays. The size or transducer footprint must fit within the window, and in extreme cases in which the transducer window is an orifice, the transducer shape must conform to the available opening. As noted above, in some applications, specialty probes, such as endorectal transducers, are needed that are small enough in diameter (size) and have the elongated shape suitable for entering a body orifice.
Second, the size or FOV and image format are selected to obtain the desired coverage over the region of interest. Here both the scan depth and image width or FOV are important. For linear arrays, the availability of trapezoidal imaging may be necessary for adequate coverage. For 3D or volumetric imaging, the extent of the image may be given as a set of maximum scan angles in orthogonal directions or a FOV and an angle. A somewhat more hidden parameter for 2D imaging for determining coverage for the region of interest is the elevation focal depth that describes the region with the thinnest slice thickness.
Third, the maximum scan depth selected determines the highest achievable frequency through the penetration relation given in the “Resolution and Penetration” section above. For example, if the scan depth is 10 cm, then, as already discussed in the “Resolution and Penetration” section, the frequency from the penetration depth d is equal to 60/d = 60/10 = 6 MHz. This frequency provides an estimate of the best lateral resolution of about 1 wavelength for an F# = 1, or, for this example, the resolution is λ = c/f = 0.25 mm (from the “Transducer Properties and Imaging” section). Exceptions to this rule are systems that use advanced signal processing to enhance sensitivity and increase penetration. In addition, the use of piezoelectric materials such as piezo composites or domain-engineered single crystals can increase sensitivity or, correspondingly, penetration depth.6
Fourth, the coverage of essential diagnostic imaging modes can be determined. From the manufacturer-provided data, the effective bandwidth needed to support different modes of interest may be extracted, or for the system considered, the actual modes of interest may be listed, such as pulsed wave Doppler, multiple imaging frequencies available, or elastographic mode. Transducers with piezoelectric materials such as piezo composites or domain-engineered single crystals can increase bandwidth substantially.6
In conclusion, transducers and image formats have evolved to better suit specific clinical applications. The classification and organization given in this article provide the background for the selection of a transducer for a particular purpose. In addition, the understanding provided can aid in determining transducer characteristics needed for new cases, thereby extending the range of transducer use.