Medical diagnostic applications and sources

ARTICLE IN PRESS

Progress in Biophysics and Molecular Biology 93 (2007) 84–110 www.elsevier.com/locate/pbiomolbio

Review

Medical diagnostic applications and sources T.A. Whittingham Regional Medical Physics Department, Newcastle General Hospital, Newcastle upon Tyne NE4 6BE, UK Available online 10 August 2006

Abstract The ways in which ultrasound is used in medical diagnosis are reviewed, with particular emphasis on the ultrasound source (probe) and implications for acoustic exposure. A brief discussion of the choice of optimum frequency for various target depths is followed by a description of the general features of diagnostic ultrasound probes, including endo-probes. The different modes of diagnostic scanning are then discussed in turn: A-mode, M-mode, B-mode, three-dimensional (3D) and 4D scanning, continuous wave (CW) Doppler, pulse-wave spectral Doppler and Doppler imaging. Under the general heading of B-mode imaging, there are individual descriptions of the principles of chirps and binary codes, B-flow, tissue harmonic imaging and ultrasound contrast agent-specific techniques. Techniques for improving image quality within the constraints of real-time operation are discussed, including write zoom, parallel beam forming, spatial compounding and multiple zone transmission focusing, along with methods for reducing slice thickness. At the end of each section there is a summarising comment on the basic features of the acoustic output and its consequences for patient safety. r 2006 Elsevier Ltd. All rights reserved. Keywords: Diagnostic ultrasound; Beam-forming; Slice thickness; Coded excitation; B-flow; Tissue harmonic imaging; Ultrasound contrast agent; Doppler; Acoustic output

Contents 1. 2. 3. 4. 5. 6.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Choice of ultrasound frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General transducer description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M-mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Types of B-mode probe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. Endoprobes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. Write zoom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4. Parallel beam forming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5. Spatial compounding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6. Multiple zone transmission focusing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.7. Developments to reduce slice thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Corresponding author. Tel.: +191 273 8811x2294; fax: +191 226 0970.

E-mail address: [email protected]. 0079-6107/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.pbiomolbio.2006.07.004

85 85 85 86 87 88 89 93 94 95 95 95 96

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

7. 8. 9. 10. 11.

6.8. Chirps and binary codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.9. B-flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.10. Tissue harmonic imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.11. Contrast-specific modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multimode scanning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3D and 4D scanning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CW Doppler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pulse-wave spectral Doppler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Doppler imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

85

. 99 100 101 102 103 104 106 107 108 110

1. Introduction This paper reviews the ways in which ultrasound is used in medical diagnosis, with particular emphasis on the ultrasound source (probe) and implications for acoustic exposure. Applications can be divided into pulseecho/imaging techniques and Doppler techniques for studying blood flow or tissue movement. The former generally use wideband pulses for good axial resolution whilst the latter generally use narrow bandwidth pulses in order to improve signal to noise ratio (SNR). 2. Choice of ultrasound frequency In order to achieve good lateral resolution, which is the ability to resolve targets lying close together, side by side at the same range, a narrow beam is required. Good axial resolution, which is the ability to resolve targets close together, one behind the other on the beam axis, requires a wide pulse bandwidth. Both are improved by using higher frequencies. Unfortunately, penetration decreases as frequency increases due to increases in both absorption and scattering. The optimum choice of frequency is therefore a compromise, being the highest that will give a useful signal from the maximum depth associated with a particular application. In practical terms this means that around 3 MHz is typical of abdominal applications in adults, around 5 MHz in children, increasing to around 10 MHz in superficial regions such as the neck or breast, around 30 MHz for the anterior chamber of the eye or intra-vascular scanning, or even 100 MHz in very superficial applications such as imaging the cornea. 3. General transducer description Nearly all medical diagnostic transducers (Fig. 1) use a thin piezoelectric disc or rectangular slab to convert electrical drive waveforms into ultrasound pulses and, conversely, ultrasound echoes into electrical echo signals. The thickness of the piezoelectric is chosen to equal half the wavelength at the required pulse centre frequency in order to achieve high sensitivity. The ceramic lead zirconate titanate (PZT) is normally used as the piezoelectric material due to its high electro-mechanical conversion efficiency. However, PZT has a high characteristic acoustic impedance compared to tissue and so, in order to avoid poor transmission across the PZT/tissue interface, a matching layer is provided. This can give 100% transmission across the interface if its thickness is one quarter of a wavelength and its characteristic impedance is the geometric mean of those of PZT and tissue. The thickness is chosen to satisfy the quarter wavelength requirement at the required centre frequency, but the transmission efficiency falls at frequencies to either side of this. In order to reduce this effect from limiting the transducer bandwidth too much, some manufacturers use two or three matching layers with progressively reducing impedances between the PZT and the tissue. This means that each interface has a smaller reflection coefficient and so frequency sensitive resonances within each layer are dampened, preserving bandwidth. Another benefit of improving transmission at the transducer face is the reduction of ‘‘ringing’’ after the electrical drive signal has ended. Such reverberations within the slab would elongate the transmitted pulses

ARTICLE IN PRESS 86

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Rear electrode

Front electrode

Lens Electrical leads

Backing layer

Matching layer

PZT slab Fig. 1. The main features of a diagnostic ultrasound probe.

and hence degrade axial resolution. Ringing is further reduced by having a backing layer behind the PZT slab. This layer has a high impedance, similar to that of PZT, to reduce reflections back into the PZT and a high absorption coefficient to absorb the ultrasound transmitted into it. A penalty is reduced sensitivity since half the acoustic power generated by the PZT is lost as heat in the backing layer. Increasingly, the original polycrystalline form of PZT is being replaced by new materials having greater electro-mechanical efficiencies and bandwidths, as well as lower impedances. One such material, composite PZT, consists of a matrix of narrow parallel columns of PZT separated by thin layers of inert low density material. Since each narrow column vibrates in a purely longitudinal mode it has greater electro-mechanical efficiency than that of a slab, whilst the low density filler material lowers the effective impedance of the matrix. Transducer elements made from single crystals of PZT have even higher efficiencies than composite PZT. Provided the acoustic matching at the faces of the transducer is efficient, higher electro-mechanical efficiency leads to larger bandwidths. This is because electrical energy is converted to ultrasound energy and removed from the transducer and its drive circuit more quickly, dampening resonance in the same way that connecting a low resistance across a resonant circuit lowers its Q value. A lens is usually incorporated between the matching layer and the patient, producing a focus in the centre of the depth range associated with the intended application. The lens material is chosen to have an impedance close to that of tissue, so the efficiency of the matching layer is unaffected. Spherical focusing can be used on single transducers that are scanned mechanically, but for linear and phased array transducers (see later) lenses must be cylindrical, providing focusing only in the plane perpendicular to the scan plane (the elevation plane). Focusing is important in determining the magnitude and position of the peak pressure and intensity produced by the probe. 4. A-mode In this mode the probe is held in acoustic contact with the patient and a stationary beam interrogates the ‘‘scan line’’ formed by the beam axis. Short pulses (typically 2 cycles) are transmitted along it and, after amplification and amplitude demodulation, the echoes are displayed as a graph of rectified amplitude versus time of arrival, which corresponds with depth (Fig. 2). The maximum pulse repetition frequency (prf) is

ARTICLE IN PRESS

Amplitude

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Transmitted

87

pulse Echoes Timebase

Range (Depth)

Fig. 2. Representation of an A-mode scan. The instantaneous rectified echo amplitude is plotted against time after transmission, which is proportional to target range for a given speed of sound.

M-mode scan

Brightness modulated scan line

Depth sweep

Scan line

Time

Fig. 3. In M-mode, a grey scale modulated version of the A-scan is swept sideways, so that each echo traces out a graph of target range (depth) versus time.

limited by the need to allow echoes from one transmission to die away before transmitting again. This limit is determined by the speed of sound and the maximum depth of interest, being typically 2 kHz or so for adult cardiac or abdominal examinations but much higher for more superficial organs such as the eye. A-mode is not available on all machines but it is useful where a precise indication of the depths or relative reflectivities of echo producing interfaces is wanted. Examples include measurement of eye dimensions prior to corneal thinning or lens replacement and characterisation of solid masses in the globe or orbit. 5. M-mode This is an extension of A-mode in which the echo amplitude is represented on the display by a grey scale and the line of echoes (time-base) on the screen is stepped sideways after each transmission–reception sequence. Echoes from stationary interfaces always arrive at the same time after transmission and so trace out a straight line on the display, whilst those from moving interfaces trace out a graph of range (depth) versus time (Fig. 3). Several seconds of M-mode may be stored digitally and instantly replayed for review. M-mode is mainly used to assess the dynamic behaviour of heart valve leaflets and heart chamber walls. An application in obstetrics is to provide a graphic proof of embryonic or fetal life by recording the cyclic heart movements. Both A-mode and M-mode involve transmitting a train of short pulses along a stationary beam. Both may be combined with B-mode (below) to help the user place the probe in the correct position. The

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

88

output then consists of a train of pulses along a stationary beam interspersed with pulses from a scanned beam. 6. B-mode This is the generic name for cross-sectional imaging achieved by sweeping (scanning) the ultrasound transmit/receive beam sideways across a plane so that many scan lines are interrogated in sequence. As in Mmode, echo amplitude is represented by a grey scale. Originally B-scanning was done by having the probe mechanically constrained by a frame above the patient such that its axis remained in a fixed scan plane while the probe was moved across the patient’s oiled skin by hand. This obsolete type of scanning is now referred to as ‘‘static’’ B-mode and has been replaced by ‘‘real-time’’ scanning. Here, the beam emanates from a compact hand-held probe held in acoustic contact with the patient. The beam is automatically, rapidly and repeatedly scanned across a scan plane defined by the position and orientation of the probe (Fig. 4). Frame rates depend on the width and depth of the scanned area, and on the spacing of the scan lines. For abdominal scanning, typically 100 scan lines might be interrogated per frame although interpolation between scan lines is used to double or quadruple the number of scan lines displayed. Frame rates for abdominal scanning are around 30 frames/s, although much higher rates are possible for superficial targets such as the eye, or where ‘‘write zoom’’ (see later) is used to reduce the scanned area. Rates can be much lower if more than one pulse is transmitted down each scan line in order to improve image quality. This is the case, for example, with multiple zone focusing, contrast pulse sequencing, coded excitation or harmonic imaging, all discussed later. B-mode real-time scanning is a major medical imaging modality, with applications in all soft tissues that do not lie behind bone or gas. It can also be combined with other techniques, such as elastography (Ophir et al., 2002), in which the elasticity throughout a tissue cross-section is estimated by applying a mechanical disturbance at the surface (commonly by pushing with the probe itself) and measuring the resulting movement of the tissue at each point in a B-mode image. The exposure received by a given point during each frame of a B-mode scan takes the form of a short sequence (the length depending on the beam width) of pulses rising and falling in amplitude as the beam passes over, with no exposure for the rest of the frame.

Probe

Chord

Fetus

Scanned beam Fig. 4. In B-mode, the beam is stepped or steered sideways within the scan plane. A grey scale modulated version of the A-scan from each line is displayed. The line structure shown here would not be visible on an actual display as the dark spaces between scan lines would be filled with interpolated grey levels.

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

89

6.1. Types of B-mode probe There are several types of B-mode probes, each with its own scanning format (Fig. 5). These have been reviewed by Whittingham (2003). Linear scans have a field of view with a constant width, making them suitable for imaging organs situated close to, or extending up to, the skin. Curvilinear scans have a diverging field of view that is as wide that of a linear array at the surface, becoming wider with depth. Sector scans have a fan shaped field of view that is very narrow at the surface, suitable for imaging wide but deep targets or for imaging through restricted acoustic windows, such as between ribs or through the anterior fontanelle of the neonatal skull. Mechanical scanning is no longer usual for the lower frequencies used for abdominal imaging, but it is the normal choice for frequencies above about 15 MHz. Mechanical scanners have a front water or oil filled compartment in which a single disc transducer is rocked or translated. Rocking gives a sector or curvilinear field of view, whilst linear reciprocation, which is practical only for the lighter, higher frequency transducers, gives a rectangular field of view. The water/oil bath has a thin acoustically transparent face which is acoustically coupled to the patient by gel. ‘‘Array probes’’, containing arrays of transducer elements, are the commonest type of probe, being compact with no moving parts, but manufacturing difficulties currently limit their use to frequencies below about 15 MHz. ‘‘Linear array’’ and ‘‘curvilinear array’’ (Fig. 6) probes contain an array of typically between 128 and 256 narrow (1.5 wavelength wide) rectangular transducer elements (Fig. 7), of which only a group of say 30 adjacent elements is used to transmit a pulse down a scan line passing through the centre of the group. The beam is stepped along the array after each transmission–reception sequence by dropping an element from one end of the group and adding one to the other. Focusing in the scan plane at a depth selected by the user is achieved by firing the outer elements of the group earlier than the central elements (Fig. 8). In reception, focusing is achieved by delaying the electronic signals from the more central elements with respect to those from the outer most elements (Fig. 9). Receive focusing is automatically and dynamically advanced to match the depth of origin of echoes (Fig. 10). The number of elements used for reception varies from say only 2 or 4 for early echoes (close targets), increasing to as many as all the elements in the array for later echoes (deeper targets). Focusing in the elevation plane (the plane perpendicular to the scan plane) is achieved by a cylindrical lens extending along the front face of the probe.

Linear

Curvilinear

Sector

"Trapezoidal"

"Compound"

Radial

Fig. 5. Types of B-mode scan formats.

ARTICLE IN PRESS 90

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 6. A curvilinear probe and the format of the scan lines.

128 leads from 128 rear electrodes

1 lead to common front electrode

Backing layer

128 elements Common front electrode

Matching layer

Cylindrical lens

Fig. 7. The basic structure of a linear array probe. A curvilinear array probe is similar apart from the array of elements being formed as an arc. A phased array probe has a similar structure, except that the width of each element, and consequently the length of the array, is typically about 1/5 that of a linear array of the same frequency.

Transmission signals at different times

Different path lengths for each element

Desired transmission focus Transducer array Fig. 8. Focusing in transmission. By having suitable differences in transmission times between outer and central elements of an array, the transmitted pulses from all the elements can be made to arrive at the desired focus simultaneously.

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Signals out of phase

Summing Signals in phase Large amplifier Delays signal

91

Different path lengths for each element

To TGC etc

Echo source at desired reception focus

Probe lead Transducer array

Fig. 9. Focusing in reception. The wavefront from a target on the beam axis will arrive at the central elements of an array earlier than at the outer elements. For a particular receive focus position, the signals from all elements can be made to arrive at the same time at a summing amplifier by having an appropriate electronic delay in each channel.

Effective receive beam

Limits of aperture expansion

F4

F29

F41

Fig. 10. Dynamic focusing in reception. The focal length is automatically increased in discrete steps by changing the receive delays. Thus, when echoes are due back from F4, delays appropriate to a focus at F4 are used. The number of elements forming the receiving aperture is also continually increased to maintain as narrow a focal beam width as possible. In this example, the receiving aperture increases with the focal length up to F29, after which it remains constant. Thereafter the focal beam width progressively increases with increasing focal length.

‘‘Phased array probes’’ scan a sector field of view (Fig. 11). They contain an array of typically 128 very narrow (0.3 wavelength wide) rectangular transducer elements, but unlike the linear array probe, all the elements are used to interrogate every scan line. The transmission beam is steered to a new angle after each transmission–reception sequence by firing the elements at different times, chosen to ensure that pulses from all elements arrive simultaneously at the transmission focus on the new line (Fig. 12). The receive focus is advanced automatically and dynamically along the new scan line after each transmission, using electronic delays to compensate for acoustic path differences (Fig. 13), in a similar way as for linear array probes. Focusing in the elevation plane is achieved by a cylindrical lens, as for a linear array probe. Modern array probes often combine both steering and stepping techniques in the same probe. An example is the ‘‘trapezoidal’’ or ‘‘virtual curvilinear’’ probe that has a field of view that is similar to that of a curvilinear probe, but without the disadvantage of a convex front face (Fig. 5). Another example is a linear array probe that offers compound scanning, as discussed later. However, beam steering with a linear array probe is more subject to grating lobe production than is the case for a phased array, due to the relatively large pitch of the transducer elements. Grating lobes are relatively weak transmit and receive beams, at angles of up to 901 on each side of the main beam, produced by diffraction when using regularly spaced arrays of transducer elements (Fig. 14).

ARTICLE IN PRESS 92

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 11. A phased array probe. An advantage of this type of probe is the small contact area required, making it suitable, for example, for viewing the infant brain via the fontanelle (right).

Transducer array Different path lengths

Transmission focus

Transmission signals at different times

Scan line

Fig. 12. Steering and focusing of the transmitted pulse in a phased array. The principle is similar to that used in a linear array, except that for all but straight ahead transmissions the focus is not situated on the principal axis of the probe.

Signals in phase

Large amplitude summed signal

Delays

S

Signals out of phase Different path lengths

Echo source at desired receive focus

Scan line

Fig. 13. Steering and focusing in reception in phased array. As for a linear array, the receive focus is automatically advanced along the scan line, although for a phased array the scan line is generally at an angle to the principal axis of the probe.

For linear array probes, the greatest value of in-water temporal average intensity (known as the ISPTA— spatial peak temporal average intensity) is generally found at the range of the elevation focus, since that is where the full output power is squeezed into the narrowest slice thickness. This dominates over the effect of the

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

93

Fig. 14. Intensity plot of the beam transmitted along one scan line by a phased array probe. A weak grating lobe is evident above the downwardly steered main beam.

operator-defined transmission focal length in the scan plane. However, the magnitude of the ISPTA does tend to increase with the depth of the scan plane transmission focus since larger groups of elements, transmitting more power, are used for greater focal lengths. For phased array and other sector scanning probes, the greatest in-water temporal average intensities are more likely to be found much closer to the probe because all transmitted beams overlap close to the probe. They are higher if the scan plane focus is set to coincide with this overlap region. In all cases, of course, the actual in situ ISPTA values are modified in position and range by attenuation in tissue, generally being weaker and closer to the probe that in-water ISPTA values. 6.2. Endoprobes These probes are designed for insertion into a body orifice or surgical wound in order to allow the transducer to be positioned very close to a particular organ. The closer proximity to the target means a higher frequency can be used, leading to higher spatial resolution. Higher pulse transmission rates can also be used, resulting in better image quality or increased frame rates. Designs are tailored to suit local anatomical restrictions (Fig. 15). For example, scanning from the oesophagus gives excellent views of the heart. Trans-oesophageal probes typically use a pair of small phased arrays mounted on a flexible tube. The two arrays are able to give mutually orthogonal cross-sections of the heart whilst being small enough not to impair the probe’s flexibility. In contrast, trans-rectal scanning of the prostate does not require a flexible probe, so that the wide proximal field of view of a linear array can be used to advantage. Intra-luminal probes operating at 30 MHz or so are passed along the inside of blood vessels in order to give unimpeded views of any pathology in the vessel wall. Most use a mechanically scanned transducer, rotating around the vessel axis like a lighthouse beam. The same 3601 scanning action can also be achieved without any mechanical scanning by using a tiny (1–3 mm diameter) cylindrical array of transducer elements (Dickinson and Kitney, 2004). This acts in the same way as a curvi-linear array but with the array being tightly curved through a full 3601.

ARTICLE IN PRESS 94

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 15. Examples of endoprobes: (a) and (b) employ curvilinear arrays and represent probes typically used for trans-vaginal scanning. (c) has a linear array and a curvilinear array arranged to allow simultaneous imaging of orthogonal cross-sections, say of a prostate transrectum; (d) uses two orthogonal phased arrays, their compactness allowing the probe to have sufficient flexibility to be inserted into the oesophagus in order to view the heart.

Despite the probe being closer, the exposure received by a target is generally no more than that associated with surface probes. This is because the total attenuation between the deepest target and the probe is likely to be similar in the two cases, the reduction in range being offset by an increase of frequency and hence of attenuation coefficient. However, if higher prfs are used, temporal average intensities can be greater. 6.3. Write zoom In contrast to ‘‘read zoom’’, which simply expands a selected region of the stored image to fill the display screen, ‘‘write-zoom’’ is a time-saving measure that restricts the scanned area when writing echo information into the machine’s image memory. The user first defines a region of interest by placing a box on the un-zoomed B-mode image. On selecting zoom mode the regions to each side of the box are left un-scanned and echoes from before and after the depth limits of the box are ignored (Fig. 16). The time saved can be used either to reduce frame rate where this is desirable, in cardiac imaging for example, or to improve image quality for a given frame rate. An example of the latter is the improvement of lateral resolution by spacing scan lines more closely or by transmitting several pulses along each scan line, each with a different transmission focus (see the discussion of multiple zone transmission focusing, later). Other ways of using time savings include the improvement of SNR ratio by using pairs of coded transmission pulses, and tissue harmonic imaging using pulse inversion (see later). A consequence of using write-zoom is that temporal average intensities are greater, since a smaller area of tissue is interrogated while the output power remains the same. This effect is greater for narrower zoom boxes.

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

95

Echoes ignored outside these range limits

Zoom Box

No scanning outside these limits Define region of interest

Zoomed image

Fig. 16. Write zoom. The user defines a ‘‘zoom box’’ which is then scanned at high resolution and/or frame rate to produce a magnified image.

Deep zoom boxes usually lead to greater intensities as well, since manufacturers often increase the number of transmitting elements when a deep focus is required. 6.4. Parallel beam forming Another important technique for saving time is to make one wide transmitted pulse serve to interrogate two or more scan lines simultaneously, each of these scan lines having its own receive beam-former (Fig. 17). The time saved by not transmitting separately along each scan line can be used either to reduce frame rate or to improve image quality, as for write-zooming. Parallel beam forming is particularly important technique for 4D imaging, discussed later. Taking an example where there are say 100 scan lines/frame and where each transmission pulse encompasses four scan lines, a given point in the tissue will be insonated by a just one full amplitude pulse, repeated every frame of 25 transmissions. This compares to the situation discussed earlier for line by line transmissions where each field point experiences a sequence of several (depending on beam-width) pulses with rising and falling amplitudes, repeated every frame of 100 transmissions. 6.5. Spatial compounding Some systems arrange for each B-mode image to be an average of several images, each obtained by sweeping the beam across the scan plane at a different angle, as shown in the ‘‘compound’’ scan format of Fig. 5. This results in better delineation of reflecting surfaces since more of a curved surface will be insonated at near perpendicular incidence. Speckle patterns and electrical and acoustic noise are also reduced by the averaging process (Fig. 18). The disadvantage is reduced temporal resolution, since the averaging creates a persistence effect. The transmissions for the several B-mode scans may not necessarily be transmitted sequentially in sets, one for each direction. Transmissions for different beams direction may be mixed up in sequence. 6.6. Multiple zone transmission focusing As mentioned earlier, in reception it is usual to vary the focal length of the transducer array with time after transmission so that, at any moment, it equals the depth of origin of any echoes reaching the transducer. This

ARTICLE IN PRESS 96

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Beamformer for receive beam A

Beamformer for receive beam B

Receive beam B

Receive beam A

Common transmission beam

Two scan lines

Fig. 17. Parallel beam-forming. Each transmission beam (between the heavy lines) is sufficiently wide to encompass two or more scan lines. A separate receive beam-former (with dynamic focusing) for each scan line selects the echoes originating on that line.

technique, known as ‘‘dynamic focusing in reception’’, greatly improves lateral resolution. Further improvement in lateral resolution by focusing the transmitted pulse is normally only possible at one range, that of the transmission focus set by the user. However, a transmission focusing technique known as ‘‘multiple zone focusing’’ allows improvements in lateral resolution at several different ranges, albeit at reduced frame rate. Each line is interrogated in several sections (zones), there being one transmission–reception sequence for each section, with the transmit focus situated at the mid-point of that section (Fig. 19). The number and positions of the several transmit foci are set by the user. A recent technique called ‘‘dynamic transmit focus’’ overcomes the frame rate penalty of multiple zone focusing by incorporating several transmission foci in one transmission (Siemens 2004). This seemingly impossible task is achieved by driving each transducer element with a voltage waveform that is the sum of several different waveforms, each of which, by itself, would result in a transmission focus at a different depth (Fig. 20). This allows several narrow high amplitude ‘‘focal’’ regions to be formed along the beam axis, but at the cost of increased acoustic noise due to broadening of the low amplitude outer parts of the beam. The amplitude and length of the pulses from each element are clearly greater with this technique, leading to greater temporal average intensities. 6.7. Developments to reduce slice thickness The techniques mentioned so far have been associated with beam forming in the scan plane. The width of the beam in the elevation plane (slice thickness) should also be as small as possible in order to eliminate echoes

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

97

Fig. 18. Compound scanning (Philips Medical). Several sweeps are combined, each with the beam at a different angle. This gives better delineation of interfaces and less noise and speckle, at the cost of some degree of persistence. Upper—cross-sectional image of a breast produced by one sweep; lower—result of adding several such images with different sweep angles.

from targets situated close to the scan plane but not actually on it. Such acoustic noise limits the ability to resolve small differences in echogenicity between different tissue masses; for small masses this may result in failure to resolve them from the surrounding tissue. Using a fixed cylindrical lens means the elevation beam width is narrow only at, or near, the depth of the lens focus. Two developments have lead to a narrow elevation beam width being achieved over a range of depths: A ‘‘1.5D’’ array probe is a limited form of two-dimensional (2D) array, having typically 5–7 rows of transducer elements, each typically containing 128 elements (Fig. 21). Focusing in the elevation plane is achieved by differences between rows in transmission times and echo delays. In reception, dynamic focusing in the elevation plane reduces slice thickness at all depths. In transmission, the focus is automatically set at the same depth as that selected by the user for the scan plane focus. The peak intensity and pressure at the transmission focus is greater than for fixed cylindrical focusing since the beam width there is a minimum in both the scan and elevation planes.

ARTICLE IN PRESS 98

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Transmission focal zone1

Transmission focal zone 2

F1

Dynamic receive focusing in zone 1

F2

Dynamic receive focusing in zone 2

Fig. 19. Multiple zone focusing in transmission. Each line is interrogated in several sections, there being one transmission- reception sequence for each section, with the transmit focus at the mid point of that section.

Waveform for F2 Waveform for F1 =

Wavefront Wavefront for F1 for F2

= =

Combined waveforms

F1

F2

Fig. 20. Dynamic transmission focusing. In this two foci example, each element is excited by the sum of two voltage waveforms, one that would, by itself, lead to a focus at F1 and one that would lead to a focus at F2. Two wavefronts are therefore transmitted simultaneously, one which converges at F1 and one which converges at F2.

Several rows of elements

Delays Effective beam in elevation Fig. 21. ‘‘1.5D’’ array. Electronically controlled focusing in the elevation plane may be achieved by having appropriate delays in the signals to and from each row of elements.

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

99

Lower frequency

Higher frequency PZT

Matching layer

Fig. 22. Hanafy lens transducer array. Each element and matching layer is narrowest at its centre (the scan plane). The central part produces a narrow higher frequency (lighter shading) beam near the transducer. The greater aperture of the outer part means that a narrow beam is also produced at greater depths, despite the lower frequency.

Another development used on some machines is the Hanafy lens. Here, the normal uniformly thick transducer elements are replaced with plano-concave elements that are thinnest at their centres, where they intersect the scan plane (Fig. 22). The thickness of the matching layer also varies, in proportion to the PZT thickness. When excited by a voltage pulse containing a range of frequencies, different parts of the transducer resonate at different frequencies, according to their distance from the scan plane. The part of the transmission emitted by the thinnest part, close to the scan plane, has the highest centre frequency. This radiation does not propagate far due to its higher attenuation, but it gives the part of the beam close to the probe a narrow elevation width due to the narrowness of the effective aperture and its high frequency. Waves transmitted from thicker parts of the transducer, further from the scan plane, have progressively lower frequencies and come to foci at progressively greater ranges from the transducer. The effective aperture is progressively wider for the thicker parts of the transducer, maintaining a narrow beam width at greater ranges. Thus the beam has an elevation width that is fairly constant over a wide depth range. The fact that the centre frequency progressively reduces with depth is not a problem since the penetration of ultrasound in tissue reduces with increasing frequency anyway. Narrowing the slice thickness by focusing in the elevation plane produces higher intensities since the same power is concentrated into a narrower area. This is not the case for the Hanafy technique since different parts of the transmitting contribute to parts of the beam at different ranges. 6.8. Chirps and binary codes A technique used to improve SNR, and hence sensitivity and penetration, is to use coded transmissions. This involves generating longer transmission pulses which carry identifying codes in the form of frequency or phase changes. One example is a ‘‘chirp’’ transmission where the pulse may contain 10 or more cycles, the frequency being swept from say half to twice the centre frequency within the pulse (Fig. 23). A matched filter in the receiver ensures that only those echoes with the same chirp waveform characteristics are passed. Another technique, known as binary coding, involves using longer pulses (e.g. 4 or 8 cycles) coded with abrupt 1801 phase changes, the inverted sections of the pulse representing the code digit 1 and the noninverted sections representing +1. The echo sequence can be decoded by autocorrelation. For a transmitted pulse with n code sections, this produces echo signals that have the length of just one code section (compression) and an amplitude n times greater (compression gain). Unfortunately the compression, and hence the recovered axial resolution, is marred by the presence of low amplitude ‘‘range lobes’’ before and after the main part of each echo signal. This problem is eliminated, albeit at the expense of frame rate, by interrogating each scan line twice. If the codes in the two transmission pulses form a ‘‘Golay pair’’, as shown in the example of Fig. 24, summing the two echo sequences after autocorrelation will cause the range lobes in each echo signal to cancel out. The summed echo sequence therefore has improved SNR, yet has similar axial resolution to that obtained when using conventional short pulses. The longer pulse lengths of chirps and coded pulses mean much more acoustic power is transmitted into the patient for a given peak intensity. Temporal average intensities and hence probe and tissue heating are therefore greater.

ARTICLE IN PRESS 100

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 23. Example of a ‘‘chirp’’ pulse. The frequency is swept within the pulse, producing echoes that can be easily distinguished from other pulses or noise.

1st pulse

2nd pulse

Fig. 24. Examples of transmission waveforms that might be used in binary coded excitation. The inverted sections of the pulse represent -1 and the non-inverted sections represent +1. In this example each ‘‘bit’’ of the code consists of two cycles; the 1st pulse is coded 1-111 and the 2nd pulse is coded 111-1. These two codes are an example of a Golay pair.

6.9. B-flow This mode combines the improved SNR arising from binary coded transmissions, as discussed in previous section, with motion detection to produce direct B-mode images of moving blood. Each line is interrogated by four pulses, in the form of two Golay pairs. After autocorrelation, the four echo sequences are summed with appropriate positive and negative weighting factors. The use of Golay pairs means that the summed signal benefits from compression gain without serious range lobes. At the same time, the weighted summation acts as a motion filter, cancelling out stationary and slow moving echoes from tissue whilst constructively summing echoes from faster moving blood. In order to show the moving blood in relation to the vessel wall and adjacent tissue, blood and tissue signals are displayed together (Fig. 25). B-flow images offer more complete filling of the vessel than do Doppler techniques as well as freedom from the angle and aliasing restrictions of Doppler (see later).

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

101

Fig. 25. One frame of a B-flow scan of a carotid artery. The faster the blood flow, the whiter the image. (GE Medical).

Fig. 26. A conventional B-mode image of a liver and gall bladder (left) and a tissue harmonic image of the same tissue cross-section (right). Note the much reduced noise in the harmonic image. (GE Medical).

6.10. Tissue harmonic imaging This mode detects and displays any second harmonic component present in the echoes, resulting in images with less acoustic noise and greater lateral resolution (Fig. 26). A second harmonic component is a consequence of pulse distortion produced by non-linear propagation, being greater for larger amplitude pulses. Such pulse distortion occurs principally in a narrow region of the transmitted beam close to the beam axis, where the amplitude is largest (Fig. 27). The amplitude of the transmitted pulse towards the edges of the

ARTICLE IN PRESS 102

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 27. Non-linear propagation produces more pulse distortion (and hence harmonic content) where the transmitted pulse amplitude is greatest. Harmonic-rich echoes are therefore mainly returned from a narrow region (shaded) near the axis, and are unlikely for reverberations or from targets in sidelobes.

beam, in the grating lobes and other sidelobes is too weak to suffer much non-linear distortion, as are reverberations and other reflected or scattered pulses. The second harmonic signal is extracted by using either a direct high-pass filter, the pulse inversion method or second harmonic-specific decoding. High-pass filtering is the original and most direct method but it gives imperfect separation of the fundamental and harmonic. Inevitably there is some overlap between an echo’s harmonic spectrum and the fundamental spectrum of the transmitted pulse. High pass filtering, therefore, causes the loss of the low frequency part of the harmonic spectrum and the passing of the high frequency part of the transmitted fundamental spectrum. The pulse inversion technique involves interrogating each line twice, with the second transmitted pulse being an inverted version of the first. Summing the two echo sequences results in echo cancellation if they are un-distorted but leaves a residual second harmonic signal if the echoes have distorted waveforms due to the transmitted pulses having suffered non-linear distortion. Chirps and binary coded pulses, mentioned above as ways of improving SNR ratio, are sometimes also used to extract the second harmonic component. Using chirps, the transmitted pulse is similar to that described earlier but, in this case, the echoes are compared to a reference chirp in which the start and finish frequencies of the frequency sweep are twice those of the transmitted pulse. Using binary coding, each line is interrogated twice as before but for harmonic detection one or more of the phase changes is 901 instead of 1801. This becomes 1801 for any second harmonic component in an echo, producing different echo codes to those of the fundamental echo component. By suitable choice of codes in the two transmitted pulses it is possible to achieve cancellation of the two echo sequences as far as the fundamental is concerned but constructive addition for the second harmonic component. Whatever the waveform of the transmitted pulse, its amplitude must be large if a useful second harmonic signal is to be detected. 6.11. Contrast-specific modes Ultrasonic contrast agents take the form of tiny (4–10 mm) spherical shells of proteins, lipids, sugars or surfactants filled with inert gas. After injection into a vein, the microspheres are carried along with the blood, increasing its echogenicity. When imaged using special contrast-specific modes (Whittingham 2005), they prove very valuable in detecting and characterising tumours (Fig. 28), clarifying the boundaries of heart chambers, measuring perfusion, and a variety of other applications. Two different approaches are used to increase the contrast between microbubbles and tissue. These are named ‘‘High MI’’ and ‘‘Low MI,’’ respectively, where MI stands for mechanical index—a pulse amplitude related safety index shown on scanner display screens to indicate the likelihood of cavitation or other nonthermal biological damage. The high MI approach involves destroying the bubbles with large amplitude pulses so that the differences in echoes before and after microbubble destruction can be detected. Pulse amplitudes

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

103

Fig. 28. B-mode image of a liver tumour without contrast (left) and a second harmonic image of the same tumour several seconds after injecting microbubble contrast agent into a vein in the patient’s arm. The concentration of bright echoes from the microbubbles around the tumour reveals an abundance of disorganised peripheral blood vessels and hence a probably malignant nature. (Philips Medical).

are not generally greater than in normal imaging, being limited in any case by FDA (Food and Drug Administration) regulations in the USA. However, the number of cycles in each pulse and the number of pulses transmitted down each line can be greater. In view of the presence of rupturing microbubbles, there is a greater likelihood of adverse biological effects. The low MI approach seeks to prevent microbubble destruction, by using low amplitude pulses, so that realtime extended scanning of the microbubble-perfused region is possible. The detection techniques used for this all make use of the very non-linear nature of scattering by microbubbles, even at low pulse amplitudes. Since microbubble echoes have a much greater second harmonic component than those from tissue, especially for low amplitude pulses, high-pass filtering, pulse inversion, chirps or binary coding, as mentioned earlier in connection with tissue harmonic imaging, can all be used. Another low MI technique, known as ‘‘amplitude modulation’’, also involves two interrogations of each line but with the second transmitted pulse having twice the amplitude of the first. Amplifying the amplitude of the first, weaker, echo sequence by two and subtracting the second, stronger, echo sequence cancels out the tissue echoes but leaves a residual for the microbubble echoes since, for them, a doubling of incident pulse amplitude does not quite produce a doubling of echo amplitude. Importantly, this residual is at the fundamental frequency of the transmission and so benefits from being at the centre of the transducer frequency response as well as suffering less attenuation in tissue than does the second harmonic. Other low MI techniques improve the rejection of echoes from moving tissue by interrogating each line 3 or more times. The successive transmitted pulses of such ‘‘contrast pulse sequences’’ (CPS) have polarities and amplitudes that differ in ways chosen to enhance or reduce fundamental and harmonic components as required. Although low MI techniques necessarily involve low amplitude pulses, the use of multiple or long pulses can mean greater temporal-average power levels. High and low MI transmissions are sometimes combined. For example, where the objective is to observe or measure reperfusion, the microbubbles in the field of view are first destroyed by transmitting a burst of high amplitude pulses. The potential for non-thermal bio-effects is clearly increased in such bursts. 7. Multimode scanning It is common to combine B-mode imaging with other modes to allow the user to check that the line or area of interrogation for the other modes is in the correct place. For example, on machines offering A- or M-mode, these traces are often displayed alongside a B-mode image, on which the relevant line or lines are highlighted. Colour Doppler images (see later) are normally displayed as a colour box overlaid on a larger B-mode image. In ‘‘duplex’’ mode, the Doppler spectrum is displayed alongside a B-mode image on which the Doppler interrogation line is shown (Fig. 34). In ‘‘triplex’’ mode, both the Doppler spectrum and a B-mode image, on which a Doppler colour image is superimposed, are displayed together.

ARTICLE IN PRESS 104

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Combining modes requires that the transmissions for B-mode are interleaved singly or in sets with those for the other mode(s). Thus, where M-mode is combined with B-mode, an adequate sampling rate for the M-mode may require one transmission along the M-mode line after every two or three B-mode transmissions. In duplex Doppler mode, however, a set of B-mode 2–3 cycle imaging transmission pulses may interrupt the stream of longer pulses transmitted along the Doppler line every second or so, or whenever the user presses an ‘‘update B-mode’’ button. 8. 3D and 4D scanning 3D (three dimensional) scanning refers to the acquisition and display of echo information from a volume of tissue, rather than from a two dimensional (2D) slice. When the scanning rate is sufficiently high to allow a real-time display of 3D information, it is called 4D scanning, the fourth dimension being time. Various methods are used to acquire echo information over a volume of tissue (Fenster et al. 2001). ‘‘Freehand’’ methods use conventional B-mode probes which are swept at right angles to the scan plane over the patient’s skin by hand. One method of continuously measuring the position and orientation of a freehand probe is by means of optical, electromagnetic or acoustic sensors or transmitters mounted on it. Some systems dispense with the need for position sensors by estimating the shift in probe position from the decorrelation of the speckle pattern in the B-mode image as the probe is moved at right angles to the scan plane. Special 3D scanning probes permit faster volume scanning and greater precision than is possible with freehand devices. These have a linear, curvilinear or phased transducer array which acquires B-mode images while being mechanically translated or rotated in an oil or water filled compartment Fig. 29). Much faster volume scanning can be achieved with ‘‘2D array’’ probes containing thousands of elements (Fig. 30). Transmission and reception beams can be electronically steered and focused in both azimuth and elevation as required anywhere within a pyramid or cone having its apex at the centre of the array. In order to make real-time volume scanning possible, one wide transmission pulse serves to interrogate a number of adjacent scan lines, each with its own receive beam-former (see ‘‘parallel beam-forming above). The volume echo data can be used in various ways. Volume rendering shows the whole tissue mass as though illuminated from behind, the brightness at each point in the image being proportional to either the

Fig. 29. Example of a 3D probe (GE Medical). The bulbous front part of the probe contains a liquid-filled compartment in which a curvilinear transducer array is mechanically swept back and forth.

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

105

2D array probe One transmission pulse serves 16 (4x4) receive lines

Fig. 30. A 2D phased array probe. In order to achieve real-time volume frame rates, each transmission beam is wide enough to interrogate a number (16 here) of adjacent scan lines. This is an extension of parallel beam forming, discussed earlier (Fig. 17).

Fig. 31. Example of a surface rendered 3D image of the face of a fetus with a foot in its mouth.

peak or integrated echo magnitude along the corresponding ‘‘ray’’ from the light source to the viewer’s eye. Alternatively, surface rendering can be used to present an opaque image of a selected organ surface as though illuminated from behind the viewer, complete with shading, colour and texture chosen to best indicate the surface 3D shape and topography. A popular application of surface rendering is to show a fetal face (Fig. 31) or body, but other applications include imaging the surfaces of structures in the heart, blood vessels and eye. Surface rendering is most easily achieved where the target surface bounds a liquid region in which the viewing

ARTICLE IN PRESS 106

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

point can be situated. However, surfaces between solid tissue masses can also be displayed using ‘‘segmentation’’ techniques, that seek to automatically define the boundaries between tissue regions with different acoustic or elastic properties. Another way of using the volume echo data is to generate cross-sectional images, or make measurements, in planes other than those originally scanned. Examples would be to obtain a coronal section of the eye or to estimate the cross-sectional area of a heart chamber. Measurements of areas in multiple parallel planes or in three orthogonal planes allow volumes to be measured, for example of the heart chambers or prostate, or in monitoring changes in tumour size. 3D scanning can reduce acoustic exposure since part of the examination time is divided into relatively brief periods of scanning alternating with periods of data manipulation and image interpretation. On the other hand, real-time 4D scanning can lead to greater exposure, especially in obstetrics where the appeal of surface rendered images of a fetus can encourage souvenir or curiosity scanning with little benefit in a risk versus benefit assessment. The use of 2D arrays can mean greater peak intensities and pressures, since the transmitted pulses are focused in both scan and elevation planes. 9. CW Doppler Continuous wave (CW) Doppler techniques are primarily used to study blood flow and for fetal monitoring. They involve transmitting continuous waves into the patient from one transducer while receiving the backscattered and reflected waves with another. Both transmitter and receiver transducers are usually housed in a single probe. They are usually either back-to-back ‘‘D’’ shapes or simple rectangles (Fig. 32a). Only moving targets that lie in the normally narrow region where the transmitter and receiver beams overlap will be detected (Fig. 32b). If an overlap region of limited depth is wanted, the two transducers may be set at an angle

T

R

T

T

R

T

R

R

T

R Cross-over region

R (a)

(b)

T (c)

Fig. 32. Probes for CW Doppler have separate transducers for transmission and reception, usually rectangular or D-shaped (a). Only targets in the cross-over region will be detected (b). Discrimination of superficial flow can be achieved by angulation of the transducers towards each other (c).

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

R

T

R

107

R R

R T

R

R R

Overlap of Transmit and Receive beams

Fig. 33. Probe for monitoring the fetal heart. This type of probe is held in place on the mother’s abdomen by a belt. In order to have a large overlap region, several wide-beamed receiving transducers surround a wide-beamed transmitter.

to each other (Fig. 32c). In fetal monitoring applications, an extensive overlap region is required in order to ensure that the fetal heart remains in both transmitter and receiver beams, however the position of the fetus changes. For this application the probe is strapped onto the mother’s abdomen. It contains a transmitter producing a wide diverging beam, and several receiving transducers, also with wide diverging beams (Fig. 33). CW transducers are made from PZT and have a matching layer and sometimes a lens, in common with pulse-echo imaging transducers. However, because the transmission is CW, they have no need of a damping backing layer. This, and the narrow bandwidth of the received signal, means that CW Doppler is more sensitive than pulse wave Doppler (below) and that very low transmitted amplitudes and intensities are normally used. The received ultrasound signal is compared to the transmitted signal and a ‘‘Doppler signal’’ is produced. This has a frequency that is equal to the Doppler frequency shift of the received signal and an amplitude that is proportional to that of the received Doppler-shifted signal. In practice, there are numerous moving targets, including both tissue interfaces and millions of blood cells, each with their own speeds and directions. This results in a complex Doppler signal having a wide range of frequencies and amplitudes in its spectrum. Doppler frequency shifts are typically a few kilohertz or less, so a simple way of presenting the Doppler signal is as an audible signal from a loudspeaker or earphones. Directional discrimination can be provided by presenting signals with positive Doppler shifts (movement or flow towards the probe) to one ear and those with negative shifts (movement or flow away) to the other. More information about the relative number and speed of blood cells in the overlap region can be obtained by displaying a real-time spectral analysis of the Doppler signal spectrum, similar to the spectrum shown in the lower half of Fig. 34. (Note that this spectrum is actually an example of one produced by pulse-wave Doppler, discussed next). In this example of a leg artery, the flow reverses direction twice in each cardiac cycle. 10. Pulse-wave spectral Doppler CW Doppler cannot discriminate between Doppler signals from different points in the overlap region, except by pre-knowledge of the characteristics of the signals to be expected from different structures or, to a limited extent, by assuming that similar structures return weaker signals the deeper they are. In contrast, pulse wave Doppler allows Doppler signals from a particular region, known as the ‘‘sample volume’’, to be isolated and assessed. In ‘‘duplex’’ mode the position of the Doppler interrogating beam is displayed as a ‘‘Doppler line’’, superimposed on a B-mode image. The user is able to place cursors along this line to define the start and end of a ‘‘range gate’’, corresponding to the near and far limits of the sample volume (Fig. 34). Pulses rather than continuous waves are transmitted, as this allows the range of the source of any given Doppler signal to be determined. It also means that the same transducer can be used for both transmission and

ARTICLE IN PRESS 108

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 34. Example of a duplex scan of a leg artery. The small B-mode image shows the user how the Doppler line and sample volume lie in relation to the artery. (Courtesy of C P Oates, Newcastle General Hospital).

reception and that this can be the same as the transducer used for imaging. The Doppler signals from a given range may be thought of as samples of the continuous Doppler signal that would have been obtained from that range if a continuous wave had been transmitted. One sample is generated for each pulse transmitted, the time after transmission of each sample being directly proportional to the range of the source. Provided the sampling rate, which is the pulse repetition frequency (prf), is high enough, a sufficiently smoothed version of the Doppler signal for frequency analysis can be recovered by low-pass filtering. The highest Doppler frequency that can be recovered (the Nyquist limit) is equal to half the prf. Only samples generated within the time interval of the range gate set by the user are passed for filtering and analysis; signals from other depths arrive when the range gate is closed and so are rejected. The pulses used for pulse wave Doppler are generally lower in frequency than those used for imaging, in order to reduce attenuation and so aid sensitivity and penetration for a given transmitted intensity. A low frequency also helps to keep Doppler frequency shifts below the Nyquist limit. In order to keep the latter high, the prf is set as high as possible within the constraint of that there be sufficient time between transmissions for echoes from the maximum depth of interest to return to the probe. A ‘‘high prf’’ mode is often available as an option for achieving a greater Nyquist limit by deliberately transmitting at two or three times the prf set by this constraint. A consequence of high prf mode is that there is ambiguity regarding the depth of origin of an echo and so the positions of all two or three possible sample volumes are presented on the display. The acoustic output in pulse–wave Doppler mode takes the form of a stream of longer pulses (3–20 cycles, depending on the gate width) at a lower frequency than is usual for imaging and as high a pulse repetition rate as possible. The long pulse lengths and high prfs mean that pulse–wave spectral Doppler is the mode that is associated with the highest temporal average intensities. 11. Doppler imaging This is a facility on some B-mode scanners which overlays a colour map on the B-mode image showing the distribution of either Doppler frequency shift or the average power of the Doppler signal. Where the colours represent average Doppler frequency shift it is called Colour Flow Imaging (CFI) or colour flow mapping

ARTICLE IN PRESS T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

109

Fig. 35. Colour flow mode (CFM) image of blood vessels in a spleen. The velocity component parallel to the scan lines (proportional to Doppler shift) is represented by the colour scale on the right. Flow towards the probe produces positive Doppler shifts (shown here as red, orange and yellow). Note the aliased region, lower left, where the yellow (Nyquist limit of +ve Doppler shift), abruptly becomes azure (Nyquist limit of ve shift). The colour box in this example nearly fills the whole B-mode image. Most applications would involve a smaller box (Philips Medical).

(CFM). Where the colours represent the average power of the Doppler signal it is called power Doppler imaging (PDI). In both cases a ‘‘colour box’’, set by the operator, defines the boundaries of the region where Doppler interrogation occurs (Fig. 35). Within the box, the tissue is interrogated as a set of 20–50 adjacent Doppler scan lines, depending on the box width. Typically, 8–12 transmission–reception interrogations are made of each Doppler line. Each Doppler line may consist of 20–50 colour pixels, depending on the box height, and for each of these there is a separate range gate to sample the output of the Doppler signal generator. Each transmission-reception interrogation therefore generates one Doppler signal sample for every pixel in the line. Thus, after 8-12 transmissions along the line, a set of 8–12 Doppler signal samples have been produced for each pixel. This is sufficient to allow an approximate estimate of average Doppler frequency and power to be made for each pixel in the Doppler line. Obtaining the Doppler information takes time and so leads to a reduction in frame rate. Smaller colour boxes must be set where low frame rates cannot be tolerated, such as in many cardiac investigations. In CFM mode, zero frequency Doppler shift is shown as black. Positive Doppler frequencies are shown typically as progressing through red and orange to yellow just before the positive Nyquist frequency limit. Negative Doppler frequencies are shown typically as progressing from dark blue through light blue to azure just before the negative Nyquist limit. (Fig. 35). Where a Doppler frequency exceeds the Nyquist limit, aliasing occurs and the colour abruptly jumps to the opposite end of the scale and advances through the scale through black again. Thus an abrupt colour change from say yellow to azure is a conspicuous indication of a Doppler shift above the positive Nyquist limit and hence of higher speed blood or tissue movement towards the probe. Conversely, an abrupt change from azure to yellow would draw the user’s attention to a higher speed away from the probe. In PDI mode, the colour scale runs from black for zero Doppler power to say orange for high power, whatever the direction of the blood flow (Fig. 36). The absence of contrasting colours makes a less noisy image, improving sensitivity to weakly reflecting or scattering targets, such as blood in narrow vessels. Flow

ARTICLE IN PRESS 110

T.A. Whittingham / Progress in Biophysics and Molecular Biology 93 (2007) 84–110

Fig. 36. Power Doppler image (PDI) of blood vessels in a neck. The colour scale, representing the power in the Doppler signal, is on the right. Note there is no indication of the direction of flow. Below is the spectral Doppler signal from a sample volume defined by the range gate. (Toshiba Medical).

perpendicular to the ultrasound beam axis may be registered, due to the non-zero Doppler shift produced towards the edges of a diverging or converging beam. (In CFM mode, the opposite Doppler shifts produced at the opposite edges of a diverging or converging beam will average to zero, so flow perpendicular to the beam axis will be shown as black). A development of PDI is ‘‘Directional PDI’’. Here say red/yellow colours are used to indicate the power of Doppler signals with positive Doppler shifts, whilst say bluish colours are used to indicate the power of Doppler signals with Doppler shifts of the opposite sign. Directional PDI thus has the high sensitivity of PDI, together with an indication of whether the flow is towards or away from the probe. The acoustic output in Doppler imaging modes typically takes the form of sets of short B-mode pulses, one pulse for each line of the B-mode image, alternating with multiple sets of around 10 longer pulses (3–9 cycles each) of lower centre frequency, one set for each Doppler line. References Dickinson, R.J., Kitney, R., 2004. Miniature ultrasonic probe construction for minimal access surgery. Phys. Med. Biol. 49 (2), 3527–3538 PII S0031-9155(04)78743-6. Fenster, A., Downey, D.B., Cardinal, H.N., 2001. Three-dimensional ultrasound imaging. Phys. Med. Biol. 46, R67–R99. Ophir, J., Alam, S.K., Garra, B.S., Kallel, F., Konofagou, E.E., Krouskop, T., Merritt, C.R.B., Righetti, R., Souchon, R., Srinivasan, S., Varghese, T., 2002. Elastography: imaging the elastic properties of soft tissues with ultrasound. J. Med Ultrasound 29, 155–171. Siemens Medical, 2004. Advancing the science of ultrasound. Siemens Medical Solutions USA Inc., P.O. Box 7002, Issaquah, WA 980277002, USA. Whittingham, T.A., 2003. Transducers and beam-forming. In: Hoskins, P.R., Thrush, A., Martin, K., Whittingham, T.A. (Eds.), Diagnostic Ultrasound: Physics and Equipment. Greenwich Medical Media, London, pp. 23–48. Whittingham, T.A., 2005. Contrast-specific imaging techniques: technical perspective. In: Quaia, E. (Ed.), Contrast Media in Ultrasonography. Springer, Bertun, pp. 43–70.