ultrasound artifact

ARTIFACTS INTRODUCTION Despite the significant technological advances in diagnostic ultrasound equipment, artifacts still represent a challenge for the sonologist and sonographer. Artifacts are echoes that appear on the image that do not correspond in location or intensity to actual interfaces in the patient. Some artifacts are undesired and interfere with interpretation. Others help to identify certain structures. It is important to appreciate that artifacts are inherent to the diagnostic method and can occur despite appropriate technique and machine settings. Only appropriate education and experience can alert the operator to the possibility of artifacts and how to recognize them (1,2,3). All methods of radiologic images have artifacts inherent to them. Nonetheless, artifacts are particularly common and can be particularly troublesome in ultrasonography. In most cases, the artifacts can be recognized for what they are. However, occasionally, an artifact can lead to a missed diagnosis or cause one to overlook a real abnormality. The physical basis of ultrasound is the key to understanding artifacts. Since the artifacts also obey the same physical principles, it may be difficult to eliminate them. It, therefore, becomes of paramount importance, to recognize them so that their significance can be appreciated and avoid potentially important diagnostic errors. Basic Assumption: Despite the greatly increased sophistication of state-of-the-art ultrasound equipment, the production of an image still relies upon basic principles. These have been extensively discussed in a previous chapter, however, to review the basic assumptions: [Fig. 1] . 1) a pulse of sound travels from the transducer in a straight line to an interface in the body and the reflected echo returns in the reverse direction back to the transducer. 2) the time taken for this "round trip" - the time of flight, is used to calculate the depth of the interface as a simple equation: D = V x T. where V, the speed of sound in biological tissue, is assumed to be constant; the same in all tissues and is set at 1540 M/sec. 3) all of the returning echoes are presumed to arise from the center of the sound beam and are, therefore, displayed along the central vector representing the beam. 4) the intensity of the displayed echo relates to the acoustic properties and size of the interface modified only by the time gain compensation (TGC). If there is any deviation from these assumptions, the machine has no way to differentiate these and will display an echo that may not correspond to the position or intensity of the original interface actually causing the echo. For example, if the beam is reflected by an interface to another outside the plane of section, the returning echo will be displayed as if it were indeed directly under the transducer. The distance will be determined by the time of flight that the sound pulse actually took. Reverberation Artifacts The first group of artifacts to discuss are those in which the sound bounces back and forth between two interfaces. This prolongs the time of flight, producing an artifact deep to the interface. The simplest of these is the reverberation artifact. In this situation, some of the sound returning to the transducer is reflected back into the patient. That pulse strikes the same interface in the patient and is reflected back to the transducer a second time. The first reverberation artifact is, therefore, twice as far from the skin surface as the original interface was from the transducer. [Fig. 2]. This phenomenon is particularly prone to occur when the sound beam is perpendicular to a strong reflector such as a soft tissue - air interface or the abdominal wall deep to a considerable depth of subcutaneous adipose tissue. Depending upon the intensity of the reflection of the interface and the degree of echogenicity of the tissue deep to the interface, one may also see a second and even third reverberation artifact. All of these will be spaced equal to the distance from the transducer to the actual interface. Generally, the reverberation artifacts caused by the abdominal wall do not cause confusion. If there is a fluid collection deep to the abdominal wall, such as the bladder, there should be no difficulty appreciating the artifactual nature of the echoes. Although there can be real echoes from sludge in the dependent portion of fluid collections, generally one appreciates the artifactual echoes on the superficial aspect of the bladder as artifacts. On the other hand, reverberation artifacts can be superimposed over the superficial portion of the liver. This gives the appearance of increased echogenicity. Although one can attempt to manipulate the TGC (time gain compensation) to balance the echogenicity in the near and far portion of the liver, it should be appreciated that the near echo may not represent true interfaces in the liver. When watching in real-time, the true echoes can be seen to move with respiration whereas the artifactual ones seem more like a haze through which the liver is being viewed. These artifacts, however, can easily obscure superficial metastases and even superficial cysts. However, if the artifactual echo appears in the right (or wrong) place, it can cause diagnostic difficulty. For example, a first reverberation artifact could be superimposed over the lateral aspect of the kidney [Fig. 4] to simulate a subcapsular hematoma. Once again, watching the kidney move during real-time scanning and moving the plane of section of the transducer should help clarify the issue. However, the hard copy image could be confusing. You cannot consider yourself a true expert in ultrasound until you have had a case in which you were mistaken, or at least perplexed, by an artifactual fluid collection simulating a real pelvic cyst or abscess. [Fig. 5]. It is actually the combination of two artifacts that simulates the fluid collection. The reverberation is generally caused by air within the rectum or other loop of bowel. The air not only reflects the sound back to cause the reverberation artifact, but also causes distal acoustic shadowing so that there will be an echo free shadow. The reverberation artifact is than superimposed on the echo free shadow to simulate the deep wall of the fluid collection. Firstly, one must always think of the possibility that the fluid collection in the pelvis could be an artifact. Attempting different angulations of the plane of section from different windows on the abdominal wall might be helpful. Partially emptying the bladder can show that the artifactual fluid collection also decreases in size. This is because the difference from the transducer to the air interface deep to the bladder decreases and, therefore, the distance from the air interface to the artifactual wall will also decrease. It is important not to empty the bladder completely lest gas filled bowel interpose and hide a real fluid collection. Some times a mirror image artifact (see next section) can cause a similar appearance so that the depth of the deep wall of the cyst may not be the same as the distance from the transducer to the rectal air. Mirror Image Artifact: The mirror image artifact is similar to the reverberation artifact. In a reverberation artifact, the sound pulse is reflected back into the body from the transducer-skin interface. In the mirror image artifact, the extra reflection comes from within the body itself. Furthermore, although the extra reflection may be within the line of the sound beam, more commonly, the sound is reflected off an angle to another interface so that like a real mirror, the artifact shows up as the virtual object. An example of this is the situation where there is a hemangioma in the liver that appears to be a lesion within the lung above the diaphragm. [Fig. 6]. The sound is reflected off the diaphragm back into the liver. The angle of reflection being equal to the angle of incidence, the sound pulse then hits the interfaces within the hemangioma to be reflected back to the diaphragm once again with an angle of reflection equal to the angle of incidence and then back to the transducer. The machine "straightens" out the path of the returning echo assuming that the interfaces were coming from the direction the transducer was pointing and at a distance corresponding to the actual time of flight. As mentioned earlier, the mirror image artifact may also simulate a pelvic fluid collection similar to the reverberation artifact. [Fig. 7]. The sound may be reflected off the rectal air at an angle so that the deep wall of the artifactual cyst represents the mirror image of the inferior and anterior walls of the bladder. In this case, it is not helpful to measure the distances from the transducer, especially in the transverse image or only a short distance of the bladder is transversed and yet the fluid collection may seem very large. [Fig. 7b]. That is because the sound is being reflected out of the plane of section to hit the inferior wall of the bladder. Mirror image artifacts can cause other strange appearances such as invasion of a transitional cell carcinoma through the bladder wall [Fig. 8]. An empyema or lung abscess can be simulated by a mirror image artifact of a hepatic cyst. [Fig. 9]. Ring Down Artifact The ring down artifact also known as the comet tail artifact, appears as a line in the direction of the sound beam and deep to a strong reflector (4-7). This artifact appears as a line in the direction of the sound beam and deep to a strong reflector. The cause can be a piece of metal in the body such as a surgical clip, or lead shot, or a BB. More commonly, the ring down artifact is seen deep to a collection of gas. [Fig. 10]. In fact, the artifact is not an echogenic line, but rather a collection of closely spaced perpendicular echoes along one or two vectors in the ultrasound image. Originally, these were thought to be sound entering the metal or gas bubble and reverberating back and forth within the structure each time sending some of the sound back to the transducer. This explanation would account for a series of closely spaced echoes equal to the depth of the piece of metal or gas bubble. However, a more plausible explanation is equivalent to the ringing of a bell. The sound pulse "insonates" the metal causing it to ring. Interestingly, the artifact does not occur deep to calcification or calculi. With gas bubbles, there is a slightly different explanation. Not all collections of gas produce ring down artifacts. However, the ring down artifact can be a characteristic "signature" of certain configurations of gas bubbles. It has been shown (7) that the "bugle" of fluid trapped between four small bubbles (the bubble tetrahedron) is the source of the artifact. The sound pulse insonates the "bugle" and causes it to vibrate and send a prolonged sound wave back to the transducer. There is a variant of this artifact that occurs deep to cholesterol crystals, usually in the wall of the gallbladder. This has been referred to as the "V-shaped artifact" (8). The characteristic appearance of these artifacts are only two or three "rings" and the more distal ones always smaller than the proximal ones. It is thought that the plate-like cholesterol crystals must be oriented perpendicular to the beam to cause the ringing (9P>enter>LORI OCT. 9/96 FINE WIR). Refraction artifact Refraction is the bending of the sound beam at an oblique interface between tissues of different acoustic velocity. There are different appearances of refractive artifacts. One type is a refractive shadow caused by defocusing of the sound beam at the edge a cyst. Another is the displacement of the distal structure by a more proximal refraction of the beam. The duplication artifact is an example where the sound beam is refracted by a more proximal interface. For example, the image of the superior mesenteric artery may be duplicated when the transducer is held in transverse plane of section over the midline. When the beam is pointing towards the right, the sound could be refracted back towards the midline by the oblique interface between the posterior aspect of the rectus muscle and the triangular fat pad behind it. This causes the SMA to be shifted towards the right. Also, when the sound beam is directed towards the left, it is also refracted back towards the midline and, therefore, displays another image of the superior mesenteric artery towards the left. [Fig. 11]. If the transducer is moved to one side or the other, the artifactual duplication can be eliminated. This type of duplication artifact could also occur in the pelvis causing a Singleton early gestational sac to appear as twins. A copper 7 IUD in the uterus can also appear duplicated. We, therefore, call this the "copper-14" artifact. Refraction can also displace the position of a deeper interface such as the diaphragm deep to a small cyst in the liver. Since the sound travels more slowly in fluid than in the liver parenchyma, a large cyst can delay the passage of sound causing the diaphragm to appear further away than it is. A small cyst, however, can cause a peculiar effect on the echo of the diaphragm. At the superior aspect of the cyst, the sound beam is refracted inferiorly to where the diaphragm is actually further way. From the inferior aspect of the cyst, the sound beam is refracted superiorly to where the diaphragm is closer. Therefore, the segment of the diaphragm can appear rotated due to refraction. Shadowing and Enhancement It is important to appreciate that the intensity of an echo is determined not only by the strength of the reflection, but even more importantly by the acoustic characteristics of the tissue between the transducer and the interface. Remember, that the sound has to go through the intervening tissue on the way down to the interface and the returning echo must also traverse the same tissue. The intervening tissue may attenuate or appear to enhance the intensity of the returning echo. Enhanced Through Transmission An increase in the amplitude of the echoes deep to a structure is called enhanced through transmission. This is a characteristic of cysts, but is actually a misnomer. The fluid only attenuates the sound less than the surrounding tissue. The cystic fluid causes no attenuation of the sound. Adjacent to the cyst, the liver parenchyma, for example, does cause some attenuation. This is compensated by the TGC making the liver echoes uniform from superficial to deep. The TGC overcompensates through the cyst causing the deeper echoes to be brighter. [Fig. 12]. The great value of this artifact is that it increases confidence in the diagnosis of a cyst. However, some times a cyst is too small for the enhancement to be explained only by the absence of attenuation. In fact, the small cyst can act as a lens, refocusing the sound beam. [Fig. 13]. This would correspond to using a magnifying glass to refocus the sun's rays to burn a hole in a dead leaf. The refocusing of the sound beam also causes enhanced through transmission. Acoustic Shadowing On the other hand, if there is acoustic shadowing (say deep to the gallstone) it is understandable that the deeper tissues may produce no echoes at all. Complete shadowing, does not cause confusion. Furthermore, it can be very helpful in confirm‎ing a gallstone or a renal stone. In fact, one should be wary of making the diagnosis of a calculus without its tell-tale shadow. There are situations in which a shadow may not be as obvious deep to a calculus. This is due to the width of the beam relative to the diameter of the calculus. If the calculus is smaller than the beam, an echo will be received back from the calculus, but sound will go around the stone to give echoes back to the deeper structures. Hence, no shadow. It is important to adjust the transmit focus to the depth of the stone to appreciate the shadow. Acoustic shadowing. Note the acoustic shadowing deep to the fetal rib. This is caused by absorption and reflection of most of the sound energy by the ossified fetal ribs [Fig. 9a]. Note the arrowhead at the right of the image indicating that the focal zone is at the level of the fetal ribs. If the focal zone is significantly deeper [Fig. 9b]. The beam is wider at the level of the ribs and the shadowing is barely perceptible since most of the sound passes by the ribs and is not blocked by them. Note in 9c, there are two focal levels, one through the level of the ribs and the other deeper. For images with multiple transmit focal distances, the vector is made up of different pulses. The pulse focus at the near level shows the shadow of the ribs down to approximately 7 cm (as in 9a). Beyond that, the shadows are not apparent (as in 9b). Focusing, obviously, plays an important role in the production of acoustic shadows. [Fig. 9c]. In clinical practice it is important to differentiate shadows due to gas-filled structures from calcified objects. Hard or calcified structures, including bone, reflect about 30% of the sound and absorbs the rest. So the shadows are relatively "clean". Gas collections reflect about 99% of the sound energy so diffuse reverberation can fill the shadow with noise making it relatively "dirty". For example, it is important to differentiate gas in the biliary tree (pneumobilia from calculi). In addition to the different characteristics of the shadows, the gas bubbles tend to have "caps" from short off axis artifacts, while the gallstones have curved proximal surfaces. It can some times be more difficult to appreciate the artifactual nature of relative shadowing. For example, if the sound beam is not completely blocked, an area of the liver may look echopoor and simulate a metastasis or focal sparing of fatty infiltration. Acoustic shadowing can also occur deep to oblique interfaces due to refraction. Refraction has already been discussed in terms of deflecting the whole of the sound beam to interfaces deep to the edge of a cyst. However, the edge of a cyst can also refract parts of the sound beam away from the other parts not traversing the edge of the cyst. The effect is defocusing. A defocused beam will have less intense sound, less intense returning echoes, and, therefore, a relative shadow. The refractive shadowing can also cause confusion if it is more diffuse and only relatively attenuating. This can occur frequently through the lower uterine segment due to the oblique interface of the superior aspect of the bladder. In the liver, the fissure for the ligamentum venosum may contain some fat and the oblique interface can cause relative shadowing of the caudate lobe simulating a metastasis. Portions of the pancreas can appear echopoor due to interposed fat deep to the left lobe of the liver. A relatively uncommon, but interesting artifact is the "two-tone testis" caused by refraction of the testicular artery. Two-tone testis. Longitudinal scan showing the more echogenic anterior superior portion and the less echogenic posterior inferior portion separated by an oblique "interface". [Fig. 10a]. Short axis view showing a narrow focal shadow indicating that the structure causing the shadow is long, but narrow. [Fig. 10b]. Color flow confirm‎ing that the cause of the refractive artifact is the testicular artery. [Fig. 10c]. Beam width artifacts Although the sound beam is displayed on the image as a very narrow vector, the beam does have a finite width. It at least several millimeters wide, in the focal zone, and even wider in the near and far zones. The intensity is higher in the center of the beam, decreasing towards the periphery. The periphery of the sound beam can be reflected by adjacent structures. So that when the center of the beam is aimed beside a strong reflector, the echo can be displayed along that vector representing the center of the beam. Side lobes of the main beam, can cause echoes to appear from structures as much as 45 degrees away from the central vector. Beam width (side lobe) artifact. Note the curved echogenic line within the bladder tangential to the bladder/uterine interface. That is when the main beam is pointing towards the cervix, the side lobe artifact can still hit the bladder/uterine interface producing a detectable off axis echo. [Fig. 11]. Side lobe artifacts are much more prone to occur with phased array and curved linear array transducers. This mechanism can cause low level echoes within the cystic object and is called beam width, slice thickness or off-axis artifact (19) It is a partial volume effect and is analogous to the artifact described for CT. This can be diminished using the correct transmit focal zone. Side and grating lobes artifacts can be diminished using the same maneuvers described for beam-width artifacts, since they are also dependent of the incident angle and non-gravity dependent. Furthermore, grating lobes can be avoided, when apodisation is used in the transducer. In this process the lateral transducer elements have less energy than central elements. Another process called spatial filtering, makes the echoes from the peripheral transducer elements less amplified, again making grating lobes less likely to occur. Dynamic focusing of the returning echo likewise tends to enhance echoes along the central axis and diminish the effects of off axis interfaces. Aside from the beam width and side lobes, a third artifact can induce the presence of low level echoes in cystic structures. This is the range-ambiguity artifact (Ref. 25 is now #21). These echoes are related to the use of fast frame rates and high pulse repetition frequency (PRF). This parameter is called "velocity scale" on many of the machines. If the depth setting is low, as for superficial structures, echoes from deeper interfaces may return to the transducer after the next pulse has been fired. This echo will be interpreted as an interface much closer to the transducer. Usually these are not noted because they are low intensity echoes that are lost in echogenic tissue. However, if there is a large cystic structure, for example, on an endovaginal scan, they can be misinterpreted as representing the far wall of the cyst. For objects situated deep in the body, or even for the dorsal skin, the time of the returning echoes can be very long, and they can reach the transducer, just after a second pulse had been emitted. The machine interpret these echoes as being originated in superficial structures. Usually they are not assigned, because they are low intense echoes compared to the time-coincident ones, from true superficial organs. However, when anechoic or low echogenicity areas are involved(gallbladder, large vessels, bladder, cyst and etc), they can be visualized as low-level echoes. Propagation speed errors As stated previously, ultrasound machines assume that the sound speed in all organic tissues is constant, about 1.504 M/sec, but actually sound velocities in organic tissues are quite variable as seen in table 3 (22). When an area containing tissues with different velocities is scanned, incorrect depth assignments are seen distally to the tissue with lower velocity. Fat (23) and fluid-filled structures are usually seen as the cause of this artifact. The most striking example is when a liver lesion with a velocity of less than 1.550 M/sec is scanned. The echoes that travel through the lesion arrive at the transducer with some delay, and are assigned in a deeper position, creating an appearance of a "break" in the structures located posteriorly to the lesion, diaphragm or even the kidney. Color Doppler artifacts As much as artifacts can be a problem with gray-scale imaging, they can cause even more trouble with color flow Doppler imaging. Color Doppler can show the presence of the flow, the direction of flow, and can give an indication of turbulent flow to help identify stenotic lesions. The newer technique of power Doppler is more sensitive at detecting flow, particularly slow flow, but is unable to demonstrate direction or velocity of flow. The artifacts can be grouped into three types: color appearing where there is no flow; no color appearing where there is flow; and the wrong color, or shade of color to confuse the direction and velocity of flow. No Color where there is Flow: Even more than gray-scale imaging, Doppler, and color flow Doppler is greatly dependent on the intensity of the sound energy reaching an interface. Since the Doppler echoes are being reflected back from moving blood cells, the interfaces are very small. They are all ?non spect to the reflectors. If the blood vessel is deep, it is very difficult to demonstrate flow. This is particularly true for large vessels like the inferior vena cava and aorta in moderate to large sized patients. Flow in the kidney can be difficult to identify unless the patient is very thin or the kidney is superficial as with a renal transplant. Transducer selection, and in fine control of the settings on the machine are essential to produce adequate color flow imaging. Firstly, one must select the appropriate transducer. Higher frequency transducers are much better at detecting the low intensity reflections from individual blood cells. However, as the intensity is attenuated greatly, lower frequency transducers have to be used. The frequency of a transducer just to demonstrate color flow Doppler is generally lower than the frequency that will suffice for gray-scale imaging. For example, the carotid artery can be visualized on gray-scale with a 10-5 MHz transducer, whereas for color flow, one requires a 5 MHz frequency. Similarly, a 3 MHz central frequency may be necessary for color flow when 5 MHz is adequate for imaging. Some machines may automatically use a lower frequency with the same transducer for the color flow pulse and the higher frequency for the gray-scale ones. It is important to use adequate gain and time gain compensation. Although this is also true for gray-scale, color flow is much more sensitive. Similarly, as will be discussed below, excessive gain or power settings will cause color noise artifacts. Frequency filters and other algorithms are used to decrease color tissue noise. However, if the filters are set too high, slow flowing blood will not show color. Color-write priority is another control that some manufacturers set automatically whereas allow the operator to set manually. It is not possible to display a color flow pixel in the same location as a gray-scale echo. The machine, therefore, has to determine which to give priority to. One can set this control so that color is only shown where the lumen is otherwise echo free. Or, the sensitivity for color flow can be prioritized so that color will show up where no vessel was visible on gray-scale imaging. [Fig. 21]. Clearly, if the color-write priority is set too low, some flow may be missed. The scale control, is probably the most important setting to adjust appropriately in color flow imaging. In gray-scale imaging, for each vector on the screen, only one pulse is necessary, however, with color flow, the data from several pulses are necessary to correctly interpret the frequencies and hence velocity. Three to eight pulses may be necessary per gray-scale pulse to adequately characterize slow flow. Faster flow can be characterized with fewer additional pulses. However, increasing the number of pulses, will decrease the pulse repetition rate for the gray-scale image and the frame rate as seen on the screen. This will cause a jerky image and the transducer must be moved very slowly. One can decrease the size of the color box on the image to regain a faster frame rate. The ability to show slow flow at a lower scale is accompanied by aliasing and inability to determine direction of flow with the higher velocities. For example, if the color scale is set appropriately to demonstrate homogeneous color in the common carotid artery, the slow flow in the periphery of the carotid bulb may not be detected and can simulate thrombus. Another example that can be frustrating is when trying to determine if there is a Budd-Chiari syndrome. In trying to determine if there are clots in the hepatic veins, it is important to have confidence that one would detect color if there was flow. This is a problem in general with color flow. The lack of color does not always mean that there is no flow. Only when slow flow can be demonstrated in adjacent structures, can one be sure that the lack of flow is significant. Color Noise - or Color where there is no Flow: Any moving structure can cause a Doppler shift. The detected Doppler shifts can be displayed on screen as color and, therefore, may simulate flow. Structures adjacent to the heart, particularly the left lobe of the liver, can exhibit artifactual color noise. Movement of the transducer can similarly cause color noise. Filters and algorithms can decrease the noise since the characteristics of moving blood cells are clearly different from interfaces which move back and forth such as in the left lobe of the liver adjacent to the heart. Nonetheless, difficulties can arise. One area that can occasionally cause trouble is a cyst in the left lobe of the liver adjacent to the heart. Color can appear in the otherwise echo free fluid collection even in the absence of flow. (Fig. 22). A special circumstance exists in tissue adjacent to a vessel with extremely high flow. This can occur around an arterial venous fistula (29). This is the color flow Doppler equivalent to a palpable thrill or an osculatory "bruit". The tissues actually do vibrate. The fistula will, therefore, appear larger than it is. On the other hand, it is easier to detect the fistula since the surrounding color noise makes it more easily detectable. Although intravenous contrast agents for enhancement of color flow Doppler imaging has not as yet caught on in routine clinical imaging, a similar artifact can occur. The stable mitral bubbles that make up the contrast agent can cause significant color noise. Dr. Peter Burns has shown that this particular type of noise can be eliminated by filtering out the main frequency so that only the harmonics of the main frequency are detected (Burns). Movement of the transducer can set up its own Doppler shift. Color noise can appear throughout the image when the transducer is moved. For example, in trying to screen for portal systemic collaterals, movement of the transducer can prevent detection of the abnormal vessels. This can be corrected by moving the transducer slower. That would limit its usefulness in searching for abnormal vessels. Some of the machine settings are important to prevent color noise. Obviously, the color gain/power must be set appropriately to minimize artifactual color. Filters and appropriate algorithms can be used to separate true flow, from other more vibratory type of motion. The color-write priority setting is very important to differentiate true flow from artifactual color. However, as was mentioned in the preceding section, the settings must be carefully adjusted so as not to miss real flow. To illustrate the problem of color noise, the Budd-Chiari syndrome case can be revisited. All of the settings must be maximized to show the slow flow of the hepatic veins, at a considerable distance from the transducer. However, this area is adjacent to the heart causing color noise artifacts. Furthermore, the patients are frequently tachypneic, which adds to the difficulties. Although color noise interferes with the detection of real flow, it is usually easy to tell that this is an artifact and not real. Because the motion is vibratory and not in one direction, the color shows random pixels of blue and red as apposed to the homogeneous color one expects in a vessel. Rarely, a cyst in the left lobe of the liver, or as in the case of Fig. 22, a dilated pancreatic duct can show artifactual color that is homogeneous (28). In these cases, it can be difficult to differentiate the artifactual color. Pulse Doppler with spectral analysis is usually helpful. A mirror image artifact is another condition that can cause color to appear where there is no flow. The principals is exactly the same for that for gray-scale imaging. For example, the subclavian vein lies on the apex of the lung. The air in the lung serves as the acoustic mirror. The sound is bounced back from the surface of the lung. The sound is reflected from the surface of the lung, to the moving blood cells and the resulting Doppler shift is reflected back to the surface of the lung and then to the transducer. The extra time taken causes the appearance of a second vessel deep to the real subclavian vessel (30, 31). [Fig. 20]. Rarely, the direction of flow can appear reversed in the mirror image artifact (Gretchen Gooding). Artifacts that Confuse Velocity or Direction of Flow: Generally, flow in the vessel is homogeneously red towards the transducer and homogeneously blue away from the transducer. During diastole, the slower flow will appear darker and in systole, the faster flow should appear brighter. There are several factors that interfere with this facile understanding. Firstly, one can reverse the colors by a switch on most machines. [Fig. 19a]. This would cause flow towards the transducer to appear blue, but away from the transducer to appear red. Furthermore, turning the transducer around can have the same effect. Although we realize on pulse Doppler imaging, one must adjust the angle of flow to appreciate the velocity, this is not possible on standard color flow imaging. Therefore, the color can relate more to the direction of flow than to the velocity. If the vessel curves on the image, where it is pointing more directly in the direction of the beam, the velocity will be perceived as faster. Where the direction is more horizontal or perpendicular to the direction of the beam, the color will be darker. Towards the centre of the vessel, the flow is faster and, therefore, the color can be brighter. Towards the periphery, the theory of laminar flow dictates that the flow is slower and, therefore, the color darker. It is assumed that the entire image is formed instantaneously. This is not true. It may take as much as a quarter of a second to produce the whole image if there is a relatively slow frame rate. Therefore, part of the color flow image may have been obtained at end diastole, and the later part in the beginning of systole. Especially in structures such as the hepatic veins, this can show two directions of flow in the same vessel! The last and most complicated effect artifact to understand is that of aliasing. This is related to the flow velocity and the pulse repetition frequency. The Nyquist frequency is to defined as two times the Doppler frequency. If the pulse repetition frequency is below the Nyquist frequency, aliasing will occur. This will cause the other color to appear where the flow is faster. If the velocity is just above the Nyquist limit, the flow will appear to be slow (dark in the other color). Even faster flow will show up as brighter in the other color. Generally, aliasing, and bright colors indicate turbulence and flow jets related to arterial stenoses or arterial venous fistulas. However, it is extremely important to appreciate the direction of flow. The same velocity flow will appear much faster and even aliased in the direction of the sound beam. [Fig. 19b]. Power Doppler Imaging: Power Doppler is a recent innovation that utilizes the same Doppler shift information in a slightly different manner. As apposed to detecting the velocity and direction of flow, power Doppler is much more sensitive in detecting the presence and volume of flow. It is important to appreciate that the brightness of a color in color Doppler is related to only the velocity not the number of blood cells at that velocity. Conversely, in power Doppler, the brightness of the color is related to the number of blood cells moving, but not the velocity. Power Doppler is even more sensitive to artifactual motion such as movement of the transducer. The frame rate is even slower than with color Doppler. However, slow flow can be detected even in small vessels such as in the periphery of the kidney or lymph nodes. Exquisite images can be made of the neonatal cerebral circulation. With power Doppler, color can appear where there are strong specular reflectors. This is different from color Doppler where the color-write priority determines that color will appear in preference where there is no echo. With power Doppler, the opposite is true. It was impressive in the earlier days of power Doppler to think that we could detect flow in the tiny vessels of the yolk sac. This turned out to be an artifact. A similar appearance can be seen in the wall of a Foley catheter! Summary: In summary, despite the continued improvement and sophistication of ultrasound technology, artifacts have not disappeared. Some of the newer transducer shapes and geometry can cause increased artifact appearances can help in certain circumstances. If an artifact appears in the wrong place at the wrong time, miss diagnoses can be made. Appropriate machine settings can eliminate only some artifacts. The best defense against miss interpretation of artifacts relies on education to recognize them for what they are.