Perspective Chapter: Quality Assurance in Diagnostic Ultrasound

Abstract

Ultrasound techniques have become a gold standard, both as a complement to conventional diagnostic methodologies and as a technique of choice. This has led to the need for quality control procedures, as happened in other sectors of diagnostic imaging. The aim of this chapter is to propose a series of physical parameters related to quality, both in B-mode and in Doppler velocimetry that could be measured with the use of commercial or “homemade” phantoms, following a protocol that considers both innovative proposals and indications published in the literature. In order to do this the different parameters are described and their physical meaning is discussed. Tests on different equipment are performed to evaluate the robustness of the protocol and the chosen parameters. The main results obtained with high clinical significance are presented. This allows both the acceptability of the equipment from a clinical ultrasound point of view and the consistency of diagnostic performance over time.

1. Introduction

Ultrasound is a very popular diagnostic imaging technique. The technology has been known since the eighteenth century, and its use in medicine was first suggested by Austrian physician Karl Theodore Dussik in 1941. Usually, however, obstetrician Ian Donald and engineer Tom Brown are cited as pioneers of ultrasound, as in the 1950s they developed a prototype of an ultrasound device, called ultrasound or sonography scanner, to be used for medical purposes, mainly to identify any fetal malformations.

Ultrasound has spread throughout the world since the 1970s. Unlike methods such as X-rays, it does not use ionising radiation but rather creates an image of the inside of the body using high-frequency mechanical waves, which cannot be picked up by our ears: ultrasound. The sounds that we can normally hear have a frequency ranging from 20 to 20,000 Hz, whilst ultrasound for medical diagnostics usually has frequencies from 1 to 20 MHz.

To carry out an ultrasound scan, the sonographer uses a small probe: this is held in hand, contains many transducers (order of magnitude 10¹, 10³) and is placed on the skin in correspondence with the volume to be examined. This device emits high-frequency sound impulses into the body, and acts both as a transmitter and a receiver, picking up the impulses reflected by the organ or tissue to be examined. More specifically, every time the sound waves encounter a separation surface between different types of tissue (i.e. acoustic impedance), part of them is reflected backwards and is picked up by the transducer; the unreflected part of the wave continues its travel.

To “seal” the space between the probe and the skin, and thus ensure better propagation of the mechanical waves, a dense water-based gel is spread on the skin immediately before the exam.

Each probe contains numerous piezoelectric crystals, i.e. ceramic or other materials that vibrate in response to the passage of electricity, producing ultrasonic waves. Vice versa, when the probe captures the waves reflected from the examined volume (echoes), the opposite happens: the transducers convert the pressure of the waves into current, and this current forms a signal which is transformed into an image by the sonographic scanner.

The instrumentation is actually able to calculate the position (depth) of the different tissues starting from the time it takes for the waves to return to the probe (being the speed of sound through the body’s tissues constant, 1540 m/s). Thanks to this information the sonographic scanner is able to reconstruct a digital tomographic image of the volume being analysed obtained by merging the several lines representing the amplitude of the single line waves received (Figure 1, real-time brightness mode, B-mode). Today, with this technology, we have the opportunity to view images in great detail [1, 2, 3, 4, 5].

Figure 1.

From ultrasound reflected echoes to pixels of a line of the image.

The frequency of the sound waves used varies depending on the type of analysis. For example, for superficial structures such as the thyroid, muscles, tendons, ligaments and breast glands, higher frequency ultrasounds with high resolution but low penetration are used (7–18 MHz), whilst to examine deeper organs such as the liver, kidneys and the heart, or to visualise the inside of the abdomen, a probe that produces lower frequency waves with low resolution, high penetration is more suitable (1–6 MHz).

Ultrasound is also used to monitor the flow in the blood vessels, through an analysis method called Doppler ultrasound or echo-Doppler. The underlying physical principle is the Doppler effect: when the receiver and the source are moving relative to each other, the sound waves perceived by the former have a different frequency than that of the sounds emitted by the latter.

In particular, if the wave received by the probe has a higher frequency than the one emitted, the receiver and source are getting closer, whilst if they are moving away, the received frequency will be lower. Doppler ultrasound is widely used in the analysis of blood flow and can provide information regarding the speed and resistance to blood flow, allowing the highlighting of vessel narrowing (stenosis) or enlargement (aneurism), increasing (narrowing, stenosis) or slowdowns (enlargement, aneurism) of the regular flow velocity.

This information can be displayed either through a velocity spectrum referring to a sampled volume (Figure 2, spectral Doppler) or by colouring the approaching and receding parts of the flow differently in the panoramic image (typically the former red and the latter blue), (Figure 3, colour Doppler).

Figure 2.

Doppler velocity spectrum as a function of time referring to a sampled volume in a vessel. Pixel brightness = blood mass. Positive speed = moving towards the probe. Negative speed = moving away the probe.

Figure 3.

Colour Doppler image. Red pixel = moving towards the probe. Blue pixel = moving away the probe. Brightness = higher speed.

The use of ultrasound techniques has become increasingly widespread in recent years, both as a complement to conventional diagnostic methodologies and as a technique of choice. This has led to an ever-increasing diffusion of ultrasound equipment, with the consequent need to apply quality control procedures on the performance of these instruments by the radiologist and medical physicist responsible of the scanner, as has happened in other sectors of diagnostic imaging.

However, it is necessary to point out an important and peculiar characteristic of ultrasound techniques: the diagnostic performance is never linked only to the characteristics of the equipment but is strongly influenced by the operator. The large number of variables involved, the numerous possibilities for different machine setups, and the direct interaction of the sonographer-probe-patient determine the impossibility of separating the quality component linked to the machine from the one linked to the operator. Therefore, the results of quality control must be interpreted in this light.

A complete quality control programme allows to:

• Make comparisons between different equipment and/or probes when choosing to purchase;

• Check the correspondence of the parameters certified by the manufacturer upon acceptance of the supply;

• Carry out status tests after each significant intervention (extraordinary maintenance, implementation of new modules, etc.);

• Regularly evaluate the constancy of the functional performance of the entire system in order to guarantee a minimum standard of diagnostic quality.

It is therefore necessary to draft a complete quality control protocol which includes, in addition to the definition of the parameters being evaluated, the identification of the limits of acceptability of these parameters. The extreme variability of the setup configurations, the different technological quality of the scanners (and cost) and the conditions of use have not allowed, at present, to agree in setting these values. We will refer to the limits proposed by the AAPM (American Association of Physicists in Medicine) working group protocol, keeping in mind that the approach commonly accepted in centres with more experience in the field of quality controls on ultrasound equipment is to set their own acceptability limits relating to the individual equipment in every foreseen configuration.

Quality checks on ultrasound equipment can be classified into clinical checks and physical checks.

Clinical checks are based on a subjective evaluation carried out by the sonographer/radiologist on a series of clinical images, which can therefore also be performed during the examination.

Physical checks are based on assessments carried out on physical parameters that can be obtained using test objects and/or phantoms (Figures 4–8).

Figure 4.

B-mode quality control phantom.

Figure 5.

B-mode quality control phantom image.

Figure 6.

Doppler quality control phantom.

Figure 7.

Doppler velocity spectrum of a Doppler phantom with laminar flow directed towards the probe.

Figure 8.

Colour Doppler image of a Doppler phantom with laminar flow directed towards the probe.

This last type of control can also be divided into two further categories: a first level, which usually produces qualitative indications, and a second, which provides quantitative information. In the first case, we proceed only with the visual investigation of the image obtained, whilst in the second level, it is necessary to have access to the numerical content of the image, downloading it to obtain numerical values of parameters. This second type of approach is advisable as it minimises errors and is also much simpler and quicker to perform. There is also software on the market that allows you to completely automate the management of second-level controls. The parameters subject to physical quality control will be listed below: each of these parameters can therefore be evaluated with a first or second-level procedure.

Clinical quality control consists of a subjective evaluation by the operator of a clinical image. The resulting judgement is obtained through the qualitative evaluation of some parameters, described below, applicable to both real-time two-dimensional ultrasound (B-mode) and Doppler velocimetry.

In this scenario and considering that sonography exams performed in 2022 as stated by the World Health Organisation were 2,8 billion worldwide (vs 3,6 billion exams using X-rays), a quality assurance programme becomes mandatory also with this technology. A set of clinical and physical parameters to be used to understand their meaning and weight has to be proposed, a test of them performed in order to assess the feasibility of their use, and some comments about acceptance limits, so different in their setting from diagnostic X-ray, discussed.

2. Diagnostic image quality (DIQ)

It is a parameter linked to the ability to represent the normal and the pathological. For this purpose, it is necessary to identify variable “targets” depending on the different clinical use (e.g. representation of the thyroid and its pathology in the evaluation of the equipment configured for the study of superficial organs; representation of superficial vessels and their pathology for the vascular study; representation of the gallbladder in the study of the upper abdomen, etc.). The DIQ is naturally linked to the next parameters.

2.1 Overall image quality (OIQ)

It is a subjective parameter that evaluates:

• The definition of the details represented (solid-liquid and solid-solid interfaces), linked to the presumed resolution capabilities of the system;

• The presence of artefacts, linked to the spatial localization capacity of the system;

• The sensation of breadth of the grey scale, linked to the dynamic range;

• Anechoic behaviour in the representation of liquid structures, linked to the signal/noise ratio of the system;

• The sensation of “real time”, linked to the frame rate allowed by the system.

2.2 Exploration efficiency (EE)

It is a parameter that depends on the dimensions of the field represented and above all on the morphological and structural characteristics of the probe (support surface, size, weight) and therefore on the ergonomics of the system. This parameter also affects the speed of execution of the exam and often the operator’s feeling of safety.

2.3 Flexibility of use (FU)

It is a related parameter:

• The amount of adjustments that are necessary when moving from one patient to another or from one probe to another;

• The simplicity of use and therefore the potential of the keyboard and measurement systems, as well as image reproduction systems. This parameter is also linked to the ergonomics of the system.

An arbitrary numerical scale can be attributed to each of these subjective parameters (for example with a score from 1 to 5: poor, sufficient, fair, good, excellent). An evaluation table of the equipment can thus be constructed for the different clinical “targets” and an overall one obtained from the sum of the partial results, with the aim of both comparing different equipment with each other and checking the stability of the characteristics over time.

3. Real-time two-dimensional ultrasound physical QC

To carry out periodic physical quality control on ultrasound equipment, it is necessary to define a minimum set of significant and measurable parameters and compare them with parameters measured upon acceptance of the machine itself, or with appropriate levels of acceptability.

To do this it is necessary to proceed with the measurement starting from a known machine state that can be reproduced in subsequent measurements. The standard configuration set for the test is called the initial setup and consists of a series of settings and adjustments to be recorded on the test protocol [6, 7, 8, 9, 10].

The parameters tested are the following:

• Accuracy and spatial linearity of measurement

• Spatial resolution

• Contrast resolution

• Uniformity

• Dead zone

• Focus

• Sensitivity (or depth of maximum penetration)

• Dynamic range

• Size, shape and filling of pseudo-cysts

• Size, shape and shadowing of solid pseudo-masses

3.1 Accuracy and spatial linearity of measurement

Most of the equipment allows the using electronic callipers, to carry out measurements of depth, dimensions of objects, areas and volumes, necessary to ensure a correct diagnosis.

Accuracy is a parameter that provides an indication of the error made in the measurement of distances and is evaluated by comparing the distance measured between two targets in the phantom with the known distance declared by the manufacturer of the phantom. By determining the relationship between the true dimensions and the measured ones, spatial linearity is obtained: this can be evaluated both horizontally and vertically. Several test objects and phantoms available on the market are equipped with a horizontal series (at a given depth) of equidistant points and a vertical series. By representing the measured distances on a Cartesian plane, on the x-axis and the true distances on the y-axis, it is possible to evaluate both the linearity of the calliper’s response (statistical p-value of regression) and its accuracy (angular coefficient of the straight line close to 1 i.e. 45°). No particular problems were found in performing this test and the results obtained were quite often excellent (p very low, angular coefficient very close to 1, typical 0.98–0.99).

3.2 Spatial resolution

Spatial resolution represents the system’s ability to distinguish point nearby objects under high contrast conditions.

The spatial resolution can be evaluated with first-level quality controls using phantoms or test objects in which point targets are placed at gradually decreasing and known distances, horizontal or vertical respectively, (e.g. 4, 3, 2, 1, 0,5 mm) and evaluated at what level the points no longer appear distinguishable in the image (first level visual inspection). These checks can be carried out at different depths and at different lateral positions with respect to the axis of symmetry of the image.

The second level of controls involves the processing of the digital image of a point-like object in an area of the field of view identified by selecting the depth and lateral position with respect to the axis of symmetry of the image. The full width at half maximum (FWHM) of the object’s two-dimensional point spread function in the axial and transverse directions is the absolute numerical estimate of the spatial resolution.

In ultrasound imaging, the spatial resolution can be axial (parallel to the direction of propagation of the beam, generally better especially for linear probes) or lateral (perpendicular to the direction of propagation, generally worse especially for convex probes). There is also a third resolution, called “elevation resolution”, which represents the possibility of separating two adjacent layers in a direction perpendicular to the displayed plane.

Axial spatial resolution is defined as the ability to distinguish two adjacent objects along the direction of propagation of the ultrasound beam. This parameter, from a physical point of view, is determined primarily by the emission frequency of the transducer. In fact, it is evident that it is impossible to resolve objects having dimensions smaller than one wavelength. For pulsed beams the axial resolution depends on the bandwidth of the ultrasound pulse: this depends on the Q factor of the transducer. The quality factor Q of the transducer is defined as the ratio between the central frequency of the transducer and its band. In particular, in two-dimensional echo imaging, transducers with low Q values are used to have a better spatial resolution, short echoes = wide band. High Q values give a very narrow band whilst low Q gives a wide band. In diagnostics it is necessary to have a wide band, in fact, this means having short spatial pulses (SPL) which allow to resolve objects very close to each other.). It must be kept in mind that ultrasound undergoes attenuation in the tissues and that this attenuation is a function of frequency, i.e. high frequencies are attenuated more, this implies a narrowing of the frequency band of the spectrum, therefore, a reduction in resolution axial as the depth of penetration into the tissue increases.

So, axial (also called longitudinal) resolution is the minimum distance that can be differentiated between two reflectors located parallel to the direction of the ultrasound beam.

$$\text{Axial resolution}=\mathrm{1/2}\times \text{spatial pulse length}.$$E1

The spatial pulse length is determined by the wavelength of the beam and the number of cycles (periods) within a pulse. Axial resolution is high when the spatial pulse length is short. Spatial pulse length is the product of the number of cycles in a pulse of ultrasound and the wavelength. Most pulses consist of two or three cycles, the number of which is determined by the damping of piezoelectric elements after excitation: high damping reduces the number of cycles in a pulse and hence shortens spatial pulse length. The wavelength of a pulse is determined by the operating frequency of the transducer; transducers of high frequency have thin piezoelectric elements that generate pulses of short wavelengths.

Supposing three cycles per pulse and using the equation c = λν, you obtain the wavelengths and the consequent theoretical maximal axial resolution values reported in the following Table 1:

Frequency (MHz)	λ (mm)	Max (theoretical) axial resolution (mm)
2.5	0.616	0.924
3.5	0.440	0.660
5	0.308	0.462
7.5	0.205	0.308
10	0.154	0.231
12	0.128	0.192
15	0.103	0.155

Table 1.

Wavelength and theoretical maximal axial resolution supposing three cycles per pulse at different frequencies.

AAPM (American Association of Physicists in Medicine) suggests the following tolerance limits on axial resolution:

• ≤ 1 mm for transducers with a central frequency greater than 4 MHz

• ≤ 2 mm for transducers with a central frequency less than 4 MHz.

The axial resolution should remain constant over time. If changes occur, the intervention of the technical assistance service is advisable.

Generally, in commercially available phantoms the pins for axial resolution are slightly moved horizontally with respect to each other in order to reduce acoustic shadow effects. Also, this test is very simple to perform but the results are very variable, depending on the depth, the position of the pins and the quality of the scanner. For example, we found resolutions close to 0,3 mm for superficial pins at 10 MHz for high-quality scanners, and significantly higher values for low-quality ones.

3.3 Lateral spatial resolution

Lateral spatial resolution is defined as the ability of the instrument to resolve two adjacent point objects (pin in the phantom) arranged perpendicular to the beam axis. It is known that in ultrasound imaging the lateral spatial resolution is worse than the axial spatial resolution. The diameter of the beam is a quantity that crucially determines the lateral resolution: in fact, a single object smaller than the width of the beam produces an echo signal during scanning for the entire time it is inside the beam, therefore the object appears as if its size were equal to the width of the beam itself at that depth, so the lateral resolution of fixed focus transducers varies greatly with depth and frequency. However, systems with multiple focal zones or dynamic focus can produce more uniform lateral resolution over a wider range of depths. The lateral resolution can also deteriorate significantly due to side lobes and grating lobes and type of probe (best with linear, less with convex, sector).

As regards the lateral resolution values and their tolerance limit, there is considerable uncertainty regarding the solution that can be proposed for routine checks. In this regard, it is useful to refer to the −20 dB pulse echo focal length, W₂₀ proposed by AAPM and proportional to the wavelength at the centre frequency and to the pulse-echo focal length and inversely proportional to the diameter of the active element (transducer).

In general, it is suggested by AAPM that the measured lateral resolutions respect the specifications reported in Table 2.

		Lateral resolution
Depth (cm)	ν (MHz)	Space between targets (mm)	FWHM (mm)
> 10	< 3.5	≤ 4	≤ 2
< 10	3.5 ≤ ν < 5	≤ 3	< 1.5
< 10	≥ 5	≤ 1.5	< 1

Table 2.

Theoretical lateral resolution for different depths and frequencies.

In real conditions (not mathematical), the optimization of spatial resolution will be obtained with the values proposed in Table 3 for a medium-quality scanner.

Frequency	Resolution (axial and lateral resolution)
MHz	Axial resolution	Lateral resolution
3.0	1.1 mm	2.8 mm
4.0	0.8 mm	1.5 mm
5.0	0.6 mm	1.2 mm
7.5	0.4 mm	1.0 mm
10	0.3 mm	1.0 mm

Table 3.

Desired axial and lateral resolution for different frequencies in real conditions.

This is for linear array transducers with parallel beams.

3.4 Contrast resolution

The contrast in the object determines a modulation of the reflected echo that returns to the probe and produces a contrast in the grey scale of the image. Contrast resolution represents the smallest contrast in the object that determines a contrast detectable by the human eye in the image, in conditions that do not have problems related to spatial resolution.

Various test objects are available with circular areas of decreasing and known contrast (scatter relative to the background). Figure 4 can be reproduced in the image. It is therefore possible to evaluate the response curve of the system (contrast in the image as a function of contrast in the object) and consequently the contrast resolution. First- and second-level checks can also be carried out for this parameter. In the first case, a visual inspection and the first insert not visible can give such an indication. In the second a plot with the grey scale contrast of the object in the image as a function of the actual one in the phantom can give a numerical assessment (contrast drop to zero). In our experience, the max contrast resolution is of the order of magnitude of 10 dB. No particular problems were found in performing this test.

3.5 Uniformity

Uniformity describes the ability of the ultrasound system to obtain a uniform image of a homogeneous object, i.e., to show echoes of the same amplitude and depth with equal brightness on the screen. Factors that influence uniformity are tissue attenuation and focus.

The attenuation of the beam in-depth, and therefore the underestimation of the echoes coming from the deeper layers, is corrected by the operator’s adjustment of the time gain compensation (TGC), which determines the amplification curve of the signals as a function of their time of delay (and therefore depending on the depth of origin of the echo itself).

By using a phantom or a test object with a homogeneous tissue texture, i.e., made up of a homogeneous set of diffusers smaller than the wavelength of the beam and correct by means of an appropriate adjustment of the TGC, a “homogeneous” image is obtained and used to calculate parameters numbers that characterise uniformity. A first-level check is only a visual inspection and, if possible. A profile of the grey level as a function of depth. One way of the second level is to select a region of interest (ROI) which is moved deeper and deeper to be able to calculate the average and standard deviation of the grey levels of the pixels as a function of the depth itself. Ideally, ultrasound images are characterised by an irregular background (texture), in which, however, the average value of the grey levels calculated in areas of interest of dimensions much larger than the “grain” of the image, should not vary either with the depth nor with the angle of displacement from the centre of the image. In reality, however, there may be areas of the image with an average grey level value different from the rest. These can manifest themselves in the form of dark vertical and/or horizontal bands: these non-uniformities can be explained in different ways. Horizontal bands may arise due to multiple focusing. In some multiple focusing systems there may be a gap between some focal zones, this can cause the beam to be wider in this zone and therefore there is a reduction in the amplitude of the echoes, this “strip” of different values of the grey is therefore completely normal because it is associated with the technological “limits” of the equipment. The presence of vertical bands in the image, however, can be caused by a damaged transducer element, but it can also be due to a defect in the transmission circuit connected to a certain element of the transducer. Vertical bands due to these electrical causes appear in the upper part of the image.

However, the absence of vertical bands in the image is not sufficient to ensure that all elements of the probe are functioning; if there is only one non-functioning element, it is very difficult to realise this in the case of phased array probes that use all their elements to create each line of the image. In this case, the absence of a single transducer in a transmission line causes a weak effect that has repercussions in all lines of sight of the image. However, it is easier to recognise such a defect in the case of vector, convex and linear array probes. In these, in fact, it manifests itself in the form of small vertical bands.

An image grey scale histogram is normally used to quantify the uniformity.

As regards the reference limits, the AAPM recommends notifying the technical assistance service if significant non-uniformities occur. The term significant is, however, subjective; each user can identify an appropriate threshold; however, a reference value of 4 dB is suggested. No particular problems were found in performing this test.

3.6 Dead zone

By dead zone, we mean the distance along the direction of propagation of the beam between the front surface of the probe and the first reflector that produces a detectable echo on the image. This region, where no information can be collected, exists because the transducer cannot emit and receive ultrasonic pulses at the same time. This depends on the instrument and is the result of both the reverberations coming from the probe-object interface under examination and the length of the pulse train. As the frequency increases, keeping the other parameters constant, the length of the pulse decreases and consequently the depth of the dead zone.

The extension of the dead zone is controlled since its variations are linked to variations in the performance of the probe and therefore of the ultrasound equipment. For example, an increase in the extension of the dead zone can be caused by a long-lasting ultrasonic pulse (caused by the fracture of a crystal), the incorrect functioning of damping materials, or the breakage of a lens. However, it must be emphasised that the reverberations observed in phantoms can differ significantly from those in patients, especially when the impedance of the phantom surface is very different from that of the skin.

Several test objects and phantoms are available for first-level dead zone control (visual inspection). If the trough between the initial pulse wave and the first echo wave is 6 dB lower than the peak of the first echo wave, then we can distinguish them and this distance is the dead zone (second level). In our experience, the dead zone ranges from 0,5 to 1,5 mm. No particular problems were found in performing this test.

3.7 Focus

The focal point of the transducer is the point of the ultrasound beam at a given depth in the medium crossed, where the intensity is maximum and the width of the beam is minimum. By focal zone, we mean the region surrounding the focal point in which the intensity of the ultrasound beam is within 3 dB of the maximum. This area is clearly the region where lateral resolution is best.

The focusing can be fixed (in mechanical sector probes) or electronic (in all others); electronic focusing can also be dynamic (multiple focal zones at the same time).

Focusing control consists of verifying the correspondence between the focal zone selected by the operator and the actual one. It is possible to verify the position of the focal zone by evaluating that of the best lateral resolution (see spatial resolution) or by using special test objects that allow direct visualisation of the shape of the beam.

Focus control can be a first-level control or a second, measuring the FWHM of the pin in the position of focus selected. We found different results depending on the frequency and quality of the scanner.

3.8 Sensitivity (or depth of maximum penetration)

Sensitivity represents the system’s ability to detect the weakest ultrasound signal that can be clearly visualised, originating from small interfaces located at a given depth in an attenuating medium. In practical terms, sensitivity represents the maximum display depth of the system under certain reproducible conditions. The weak signal in question is detected in the presence of noise, and its detectability is an indication of the signal-to-noise ratio. The factors that can influence the sensitivity are the frequency and intensity of the excitation pulse; the gain; the TGC; the focusing; the attenuation of the medium; the depth; the composition and geometry of the reflecting element; and finally, the noise due to the system electronics. The signal/noise ratio is maximum within the focal area and decreases on the sides of it.

By carrying out this type of check, it is necessary to be able to distinguish the echoes diffused from the background from those generated by electronic noise. The former, by keeping the transducer stationary on the phantom, will appear stationary, whilst the latter will present fluctuations.

For this parameter, it is possible to carry out first and second-level assessments.

In the first case, a vertical inspection of the pin in the phantom is performed and the depth of the last visible pin is measured. In the second a rectangular ROI around the pins is acquired and its profile analysed. No particular problems were found in performing this test. The depth of penetration (sensitivity) is generally limited to approximately 200 wavelengths, corresponding to a depth of 30 cm for a 1 MHz transducer, 12 cm for 2.5 MHz transducer, and 6 cm for a 5 MHz transducer.

3.9 Dynamic range

The dynamic range of an ultrasound scanner is defined as the ratio, expressed in dB, between the intensity of the largest echo and that of the smallest echo, which can be displayed simultaneously. The smallest echo is the one that is barely distinguishable from the system noise, whilst the largest is the one just below the saturation level. Therefore, the dynamic range provides a measure of the intensity (or amplitude) that the system can handle. All elements of the equipment, from the probe to the screen, influence the dynamic range. The dynamics of the signals reaching the probe are normally much wider than that which can be represented with a normal screen; a non-linear compression (usually logarithmic) of the intensity of the echoes is therefore necessary. In particular, most of the grey level range is reserved for weak signals, compressing very intense signals more.

The dynamic range can be evaluated with first and second-level controls with phantoms or test objects similar to those used for contrast resolution. No particular problems were found in performing this test. The typical value we found can reach 100 dB.

3.10 Size, shape and filling of pseudo-cysts

Cysts are fluid-filled, hypoechoic and normally poorly attenuating structures. The shape, size and consistency of the cyst must be represented correctly.

The first level of control of these parameters allows you to evaluate any changes in the system output, an incorrect selection of the TGC curve, an incorrect insonation non-uniformity, an incorrect focusing or an insufficient quantity of number of lines of sight in the image.

Various phantoms suitable for this purpose are available for the evaluation of the circular shape of the circular insert in the phantom. No limits are suggested.

No particular problems were found in performing this test.

3.11 Size, shape and shadowing of solid pseudo-masses

Solid masses are generally echogenic and normally with high attenuation. The shape, size and consistency of solid masses must be represented correctly.

The first level of control of these parameters allows you to evaluate any changes in the system output, an incorrect selection of the TGC curve that leads to an incorrect uniformity of the image, an incorrect focusing or an insufficient quantity of number of lines of sight in the image.

The result of the test is correct if all the scanner parameters are optimised and particular attention to the angles is taken.

4. Doppler velocity QC

To carry out periodic quality control on ultrasound equipment used as a Doppler velocity metre (continuous Doppler, pulsed, colour, power, etc.), it is necessary to define a minimum set of significant and measurable parameters to compare with parameters measured at the time of acceptance of the machine itself, or with levels of acceptability. To do this it is necessary to proceed with the measurement starting from a known machine state that can be reproduced in subsequent measurements. The standard configuration set for the test will be called the initial setup and consists of a series of settings and adjustments that must be recorded in the test protocol.

Some parameters, although important for image quality, are intrinsically determined by the setup of the equipment, such as the minimum measurable speed, determined by the choice of the wall filter once the angle has been fixed Doppler, the maximum measurable speed, determined by the choice of PRF, once the Doppler angle has been fixed and so on.

Therefore, no quality control is carried out on these parameters.

The parameters tested are the following:

• Flow sensitivity

• Accuracy of speed estimation

• Precision of speed estimation

• Accuracy of positioning of the sample volume

• Spatial resolution in colour

• Colour contrast resolution

• Coloured noise

• Flow registration

• Temporal resolution

• Motion discrimination

4.1 Flow sensitivity

It represents the minimum flow that produces a statistically different signal from the background and can be variable depending on the average velocity in the vessel. This parameter is measured by simulating a known, gradually decreasing flow in a phantom shown in Figures 6–8 (in this case it is necessary to use synthetic blood), or in a test object (preferably decreasing the diameter of the pseudo-vessel and keeping the average speed constant). In fact, as is known, the laminar flow in a duct at a given instant is equal to the product of its section at a point and the average velocity of the particles passing through that section at that instant. It should be remembered that in laminar conditions, the speeds of the particles close to the axis of the vessel are greater than those placed near the walls and that this difference increases as the diameter of the duct decreases. This first level test is performed with the selection of the lowest wall filter and decreasing the velocity until the flow disappears in the image (spectral or Doppler).

In our experience, the lowest detectable velocity was 0.04–0.06 cm/sec in a 3.00-mm tube. Detection was about three to four times less sensitive in a 0.30-mm and 0.05-mm tubes. Amplitude Doppler US sensitivity is only slightly dependent on the angle of incidence and propagation medium.

4.2 Accuracy of speed estimation

It is obtained by comparing the true (known) average speed inside a pseudo-vessel with the one measured by the velocity meter, for the different speeds of interest.

The parameter in question depends, for a given flow, on the velocity profile inside the duct itself. If we assume that the flow is laminar, and therefore the velocity profile is parabolic, it is possible to estimate the peak velocity v_p using the following relationship:

$$v_p=2\overline{v}$$E2

If the flow inside the vessel is turbulent, or has characteristics other than laminar ones, estimating the peak velocity is more problematic.

Generally, when a liquid enters a tube it does not present a laminar regime but reaches it after a certain distance whose length depends on the diameter of the tube, the speed and the viscosity of the liquid. In commercially available phantoms there are generally some sections of the pseudo-vessels in which the laminar regime conditions can be assumed to be satisfied. It is therefore necessary to measure the speed within these zones and compare it with that calculated with the following expression:

$$\overline{v}\left(cm/s\right)=\frac{\phi \left(ml/s\right)}{A\left(c{m}^2\right)}$$E3

where φ is the flow rate and A is the vessel cross-sectional area.

In our experience, Doppler methods are capable of good absolute accuracy when suitably designed equipment is used in appropriate situations, with systematic errors of 6% or less (Figure 7, steady flow). This is a second-level test.

Particular attention must be taken in order to measure the parabolic region of the flow.

4.3 Precision of speed estimation

It consists of checking the reproducibility of a speed measurement and is obtained by carrying out subsequent measurements in the same conditions and checking the constancy of the measured value. The precision, taking attention at an angle is generally high. Particular attention must be taken in order to measure the parabolic region of the flow.

4.4 Positioning accuracy of the sample volume

The sample volume delimits the region of the image, and therefore of the object, from which Doppler signals are obtained. The check is carried out by positioning, at different points inside the vessel, the smallest sample volume available and carrying out a flow measurement in points from one wall to the opposite one. The highest speed value must be obtained when the sample volume is found at the centre of the vessel. It is a check equivalent to that of the accuracy of the speed estimate: a misalignment produces a variation in the average speed measured.

The results we obtained were very spread due to the quality of the scanners.

The test requires particular precision.

4.5 Spatial resolution in colour

It represents the system’s ability to spatially discriminate two different flows, in conditions of high contrast, for example, two opposite flows, when they are very close to each other.

Typically, it can be checked by simulating two pseudo-vessels of decreasing diameter and increasingly closer (and parallel) to each other, in which two opposite flows flow (therefore one represented in red and one in blue).

It is necessary to evaluate the minimum conditions of flow velocity and distance that produce an image in which the flows begin to be unresolved (Figures 9–12).

Figure 9.

Scheme of the homemade Doppler phantom for colour resolution.

Figure 10.

Homemade Doppler phantom for colour resolution.

Figure 11.

Doppler phantom image with colour resolved.

Figure 12.

Doppler phantom image with colour not resolved.

If the phantom does not have two adjacent pseudo-vessels, it is still possible to carry out the check using a flow perpendicular to the ultrasound beam. In this case, the Doppler signal is affected by spectral broadening. If the flow is orthogonal to the beam, the broadening artefact must be the same for both positive and negative speeds and the amplitude of the spectrum must be the same. If the velocity spectrum was obtained in pulsed Doppler mode it must be symmetrical with respect to the axis. The values of vessel diameter and flow velocity that produce a flow not resolved are an index of the second level of spatial resolution in colour. Value obtained is very spread.

4.6 Colour contrast resolution

It represents the system’s ability to discriminate different speeds, even in the same flow.

Typically, it can be checked by evaluating what is the minimum difference between two known speeds that determine two distinguishable speeds in the image. It is a second-level test. Value obtained is very spread.

4.7 Colourful noise

Coloured noise consists of the detection of flow in areas where there is no flow. Typically, it is sufficient to acquire a colour image using a phantom or a test object in conditions of no flow and gradually increase the colour gain until coloured noise appears. The higher the gain at which the coloured noise appears, the better the performance of the ultrasound. It is a second-level test. Value obtained is very spread.

4.8 Flow registration

It represents the system’s ability to give a colour signal exactly superimposed on the morphology of the pseudo-vessel. The check consists of carrying out a scan of a vessel using the Colour Doppler mode and verifying that the colour information on the flow is contained within the vessel, does not overflow but does not leave areas with an absence of colours, especially near the walls. It is a first-level test. Value obtained is very spread.

Particular attention must be taken to the overlap colour-B-mode.

4.9 Temporal resolution

It is related to the colour frame rate of the image. A dynamic event is simulated: speeds change rapidly and acceptable reproduction of dynamics in the image is verified. It’s a first-level test.

4.10 Motion discrimination

It represents the system’s ability to discriminate between the movement of liquid and solid masses. It is influenced by the choice of wall filter.

It is checked by simulating a known flow in a pseudo-vessel with very elastic and extensible walls and evaluating the signal of the walls, which appear displayed and which must not be in colour. It’s a second-level test.

Referring to the frequency of evaluations, there are no indications whatsoever from international bodies or centres with particular experience.

5. Conclusions

It should be noted that a complete quality control programme requires some hours of machine time for acquisition only.

However, due to the very high number of ultrasound exams performed worldwide, we recommend at least a complete annual check for each piece of equipment, reserving the right to decide to carry out any simple checks at higher frequencies, which require little waste of time, and in any case whenever the radiologist deems it necessary.

Planning a quality control programme can reduce the number of repeated investigations, make the diagnosis more accurate and limit patient referral to other methods for diagnostic completion, making each ultrasound examination more accurate and informative.

It concerns real-time B-mode ultrasound and Doppler velocimetry, both spectral and colour mode.

It should follow a protocol that contains the parameters to be evaluated, the phantoms to be used, the measurement methods, the frequencies of the evaluations and, possibly, the tolerance limits.

A set of controls both for B-mode and Doppler are proposed. The parameters investigated were chosen following their clinical relevance and in the cases was possible, limits of tolerance suggested.

The tests were all used in commercial or homemade phantoms and the problems and general results were discussed. This can be a simple way to approach the physics of quality control in ultrasound. If someone wants to go into detail some useful reference is suggested. With the very high number of exams worldwide performed, it is mandatory to follow a QA programme like in the case of X-rays.