CEUS peculiarities, pitfalls, tips and tricks

Elaboration on CEUS based on our experience. If your CEUS exam is not conclusive, knowing about these CEUS peculiarities and pitfalls and using optimal exam technique may hopefully help leading to a more convincing conclusion.

Red links in the guidelines open live video samples. These files are large, since they are tuned for diagnostic quality. In some cases there is also a low resolution alternative. 

Characterization

Haemangioma

FNH

Metastasis

HCC

Regenerative nodule

Skip lesion/fatty infiltration

Abscess

Adenoma

Detection

Metastases

General pitfalls

Over the years we have encountered many technical issues, particularly during the first year of CEUS work. The past years we have also had the honour of being consulted many times about tricky CEUS exams that have been sent to us. In some cases lesions certainly do not have circulatory patterns that are very typical, and a biopsy may be the only answer. However, our own observations and some of the consultations have made us aware of eight conditions that may be to blame in inconclusive cases where the pathology is otherwise straightforward:

Mechanical Index mistakes

Image noise

Too persistent scanning

Lesion too deep

Incorrect focal points

Earliest phase not seen

Reflection or "glare"

Low frame rate

Characterization

Haemangioma Up

The typical haemangioma shows initial spots or "puddles" of intense enhancement in some or all of the periphery while none more centrally, followed by a continuous growth of these puddles of enhancement toward the centre, and there is no washout in the late phase. The propagating enhancement pattern from the periphery to the centre may take anywhere from a few seconds up to several minutes, but should nonetheless be readily recognizable for the diagnosis. Within the first 30 seconds following TCA it is usually quite clear whether the lesion is probably a haemangioma or something else. The early presumption of a haemangioma may be a reason to minimize MI and extend scanning intervals.

Non-enhancing central areas are quite frequent, but the general character of the arterial enhancement pattern should be recognizable for dependable verification of a haemangioma. Other lesions than haemangiomas show immediate or almost immediate enhancement in all vascular areas, albeit being very different from one another in vascular architecture and degree of vascularity. Care must be taken not to mistake hypovascular or necrotic areas of [metastases] for the initially non-enhancing parts of haemangiomas.
Typical haemangioma
with a small non-enhancing centre.

Fast-flow haemangiomas should have the same basic characteristics for diagnosis as more common haemangiomas despite the fact that they enhance from the periphery toward the centre in a matter of seconds.
Fast flow haemangioma
Enhances completely within ten seconds. No washout.

Non-enhancing haemangiomas are encountered occationally. Some small lesions looking like bright haemangiomas on baseline have not enhanced whatsoever, and some have been biopsied confirming haemangiomas. These are probably completely thrombosed haemangiomas, and we tend to report them as such in otherwise healthy individuals.
Non-enhancing haemangioma
, characterized as such by being completely avascular, probably thrombosed.

Haemangiomas with true washout are encountered once in a while. Such washout is generally seen later than that of a typical metastasis. However, clear visualization of convincing bright initial peripheral enhancement with the mandatory central propagation is a very reliable sign of a haemangioma. With experience a haemangioma showing washout is not necessarily a diagnostic problem if the early phases of the exam are unequivocal. This is a good example of when the excellent time resolution of ultrasound is an asset over that of other modalities. []

MI settings and scanning time within "normal CEUS limits" may occasionally be too much for haemangiomas. In most haemangiomas the circulation is extremely slow, a reason why the virtually stationary microbubbles seem more susceptible to mechanical destruction by sound waves than in other lesions. Under unfortunate circumstances this means that a "false washout" may be inflicted in superficially located haemangiomas by the examiner by exposing the haemangioma to an otherwise "normal CEUS level" MI, while the enhancement of the surrounding liver is actually preserved. The haemangioma may even falsely seem not to enhance at all. While a re-injection and a new exam with a very low MI setting may lead to a conclusion, it is advisable to scan superficial lesions cautiously according to the Sonoexam characterization protocol in the first place, in order to avoid misunderstandings.
Iatrogenic bubble destruction in haemangioma
(low res) Examiner inflicted darkening of shallow haemangioma in late phase, by going from MI 0,07 to 0,18 (which normally is within normal CEUS limits) and prolonged examination time.
Enhancement of haemangioma not seen
(low res) Microbubbles in slowly enhancing haemangioma are destroyed as they enter the lesion. Remedy is re-injection at a very low MI and finally long intervals between Sonoscans.

FNH Up

The typical FNH has a typical arterial enhancement from the centre and out, "spoke wheel" or tree shaped arteries and a non-enhancing central scar. Typically there is no washout of contrast in the late phase, thus indicating the benign nature of the FNH.
Typical FNH
Large FNH with all FNH characteristics present.

Roughly 50% of FNH's lack one or more FNH criterion except for the late phase enhancement, or sometimes some of them can not be seen unless settings are optimized. This means that many FNH's show two or less arterial phase characteristics and some show none. Typically there is a feeding artery which initially reaches directly into the centre of the FNH, and the arteries of the FNH enhance very quickly (within a second) from this centre and out. The central scar is obvious in all exam phases. If a lesion does not reveal any typical FNH characteristics we rely on the absence of washout in the late phase only in otherwise healthy individuals without a history of cirrhosis or hepatitis.
Uncharacteristic FNH
Diagnosed as highly probable FNH on the basis of no washout in healthy individual. Small haemangioma is also seen.

Arterial center-to-periphery flow may be visualized as a directional flow of microbubbles before the entire FNH is uniformly enhancing, provided the framerate and region of interest are optimized. An optimal exam should aim at distinguishing an FNH from other lesions by displaying the special vessel characteristics.
Arterial center-to-periphery flow
same as Typical FNH above, with slow motion workup at workstation.

A "fire ball" of tissue enhancement from the center to the periphery of the lesion may occationally be captured by the Sonoscan although the arteries are not clearly visualized. A sufficient frame rate along with a proper region of interest is a prerequisite as well as slow motion reading at a monitor. When present, the "fire ball" may be a useful aid, but the pattern is a sibling of the "basket sign" of some HCC's so it is wise to balance it against clinical and other parameters. Nevertheless, in our experience a conspicuous "fire ball" pattern may be diagnostic for an FNH.
FNH, "fire ball"
Slow motion workup at workstation

A central scar is usually straightforward.

FNH's with a clear washout have been encountered, although they are rare. Washout probably depends on degeneration. If the arterial characteristics and the central scar are clearly recognized, the lesion may still be diagnosed as a highly probable FNH, but great care and convincing documentation is necessary in order not to overlook a malignancy. Frequently FNH's with washout lead to biopsy.
[FNH with washout] Biopsy verified FNH with no arterial phase FNH characteristics and washout

Further tips on exam techniques for quickly enhancing FLL's

Metastasis Up

The typical metastsis is characterized by quite rapid washout in the portal or late phase, usually visible within 30-60 seconds after TCA and conspicuous after 90-120 seconds. All attempts to distinguish one type of malignancy from another by the appearance of the metastasis circulation have been disappointing in the light of a known primary tumour or subsequent biopsies.

A characteristic "rim zone" appears as a hypervascular rim around or in the circumference of many metastases, but far from all. The rim zone varies from thin to thick, but is usually quite uniform in the individual metastasis. It may represent microinfiltration or inflammation in the adjacent liver or a hypervascular surface of the metastasis itself.
Thin rim zone in an intermediatly vascularized metastasis.

The arterial phase appearance varies greatly, from very hypovascular via "isovascular" to hypervascular. The hypovascular metastases are hypoechoic to the liver in all phases, but show washout and become even darker after the arterial phase.
Hypervascular metastasis with rapid washout.
"Isovascular" metastasis practically "isovascular" with liver in arterial phase, quick washout.

A very hypovascular lesion is obviously very different clinically from one which is truly avascular, such as a simple cyst, thrombosed haemangioma or RF-ablation. CEUS is extremely sensitive even to minute enhancement, but detection of enhancement in very hypovascular lesions requires careful optimization of the machine settings, especially regarding CEUS noise level. With optimal settings CEUS can often rule out enhancement with quite great confidence.
Hypovascular metastasis hypoechoic to liver in all phases. Note the thick rim zone. Also metastasis with large necrosis and haemangioma.

Necrosis appears as non-enhancing areas, usually in the centre with an irregular margin to viable tissue, but there is great variation in the shapes, sizes and numbers of necrotic areas. Very few metastasis are so necrotic that viable tissue is undetectable. Superficially hypervascular, very necrotic metastases may at first glance mimic a haemangioma, but all parts of metastases that do enhance typically show washout. Such a rim of superficial washout next to necrosis is not as conspicuous as the preceding arterial phase enhancement, and may be at risk of being overlooked. Also, the rapid flow of UCA is often seen in the Sonoscan. 
Metastasis, large necrosis Almost complete washout in the viable superficial tissue.

Absent washout is extremely rare, but we have seen a few cases of metastases from renal and uterus cancer which display very slight washout, and thereby being difficult to detect with CEUS. However, these cases comprise a mere handful of all our cases since 2002, and are very uncommon indeed. []

HCC Up

HCC, or Hepatocellular Cancer, can be all from well circumscribed solitary lesions to diffusely spread and practically infiltrating the liver. In larger lesions there are frequently irregular areas of necrosis. HCC's are common in liver cirrhosis, especially on the basis of hepatitis, but quite rare in otherwise healthy individuals.

Rapid, intense hypervascularity is typical for HCC's. Enhancement prior to surrounding liver also in cirrhotic livers, which themselves enhance earlier than normal because of arterialization of the parenchyma following reduced portal vein flow. [multifocal]

No or little washout is unfortunately the case in many HCC's. Clear washout is a sign of malignancy, but the absence of washout is of no real significance. This is the reason why HCC always should be suspected in the advent of a hypervascular lesion in a liver with cirrhosis or hepatitis, unless it is clearly a haemangioma or an FNH. []

The arterial phase enhancement pattern is often unspecific, but for some lesions a detailed study of the arterial phase reveals patterns that support an HCC:

Curly, irregular arteries can sometimes be identified in the HCC in the arterial phase. []

The "basket sign" is a term for the visualization of rapid enhancement from the entry point of a feeding artery on one side of the lesion to the other, combined with a characteristic finger-like pattern of arteries that originate from the same spot and continue inside and near the surface of the lesion. In essence, the distinction between the typical basket sign and the typical FNH "fire ball" enhancement is that the basket sign can be described as is "fire storm" of enhancement in one direction along with arteries bulging along the lesions periphery, while the FNH has its main arteries more centrally, from which the enhancement expands to the periphery. In the individual case a sound scepticism is recommended before reading too much into this delicate distinction, but if clear and conspicuous the basket sign can indeed be diagnostic for HCC. Again, these arterial specifics are seen in the very first seconds following TCA, and may require slow motion scrutiny.
HCC, basket sign with slow motion at workstation. Diagnostic for HCC. Surgical resection performed on CEUS characterization. Washout is not reliably strong.

Regenerative nodules in cirrhosis Up

Courtesy of Dr. Anna-Karin Siösteen-Tofte, Karolinska University Hospital, Sweden

In the arterial phase regenerative nodules may appear as virtually black lesions compared to the enhanced liver until the portal enhancement of the lesions begins. They differ from most other FLL:s (except haemangiomas) in that they enhance later that the surrounding cirrhotic liver parenchyma. It is speculated that the regenerative nodules have a more normal circulation than the cirrhotic liver, which means that they depend more upon the portal than the arterial flow for their own blood supply. The cirrhotic liver has much poorer portal flow, which is compensated by increased arterial flow. Thus, the UCA bolus enhances the cirrhotic liver prior to the regenerative nodules.

In the portal phase the enhancement of the lesions is quite rapid and uniform, and there is no difficulty differentiating them from haemangiomas.

There is typically no washout in regenerative nodules. Once they are isoechoic with the rest of the liver, they can no longer be seen with CEUS.

In HCC detection the characteristics of regenerative nodules make them quite conspicuously dark in the arterial phase, which means that they turn up as black lesions while scanning in the arterial phase to detect HCC:s, but they quickly "disappear" following TCA and may be isoenhancing by the end of the arterial Sonoscan.
Regenerative nodules in HCC detection Appearance of regenerative nodules in cirrhotic liver. Note that nodules are not seen at the end of the scan, since they have enhanced by then.
Courtesy of Dr. Anna-Karin Siösteen-Tofte, Karolinska University Hospital, Sweden

Dysplastic or malignant transformation make the regenerative nodules enhance quicker that the surrounding liver cirrhosis. It has been encountered that a defined lesion has turned from "early dark" in one exam to "early bright" half a year later, as it has progressed from regenerative to dysplastic or malignant. []

Skip lesion in steatosis/fatty infiltration Up

Fatty lesions enhance like the liver. This fact is actually the basis for their characterization; there is no difference in comparison to the surrounding liver parenchyma in any of the circulatory phases. The vessel architecture, speed of enhancement and late phase enhancement are typically identical to that of the surroundings. Sometimes a minute echo difference can be seen in the late phase, but usually the lesion is discretely hyperechoic in these cases, which very rarely causes diagnostic problems. We have not seen any exceptions to these characteristics so far. [fatty]
Skip lesion Enhancement pattern like the surrounding liver parenchyma, slightly brighter in late phase

Abscess Up

The echo of abscesses may be very isoechoic. While most larger abscesses are readily seen on conventional ultrasound, they may be notoriously evasive, often discrete and may easily be overlooked. If one is discovered there may be many more that are not seen because of different echo. On the other hand,

CEUS on liquefied abscesses makes them very conspicuous. Abscesses should liberally be subject to CEUS detection according to the same principles as metastases mentioned above.

The inflammatory capsule is typically hypervascular, and thicker capsule areas typically wash out. However, not all abscesses show a clear capsule, probably depending on their age and the aggressiveness of the pathogenes involved. There are also abscesses that appear as simple cysts on US, but these may be characterized by the very thin hyperenhancing periphery in the arterial phase. This appearance may be brief, why CEUS parameters must be optimal. The inflammation of their periphery may be far to thin to be seen as washout in the late phase.
Liver abscess slightly hyperemic capsule, and with some incomplete septations.

Septations within the abscess are often surprisingly enhancing due to CEUS's extreme sensitivity to small amounts of UCA. Thin septations may be exaggerated on CEUS, and not necessarily a sign that the abscess is not "ripe" for drainage. []

Non-liquefied abscesses resemble metastases (such as many fungal abscesses or early bacterial abscesses). They are typically hypervascular and show a washout in the late phase; in fact, such abscesses fulfil the criteria of metastases, and the same detection technique is used.
Fungal abscesses in patient with leucemia.

Adenoma Up

Adenomas are quite rare, and there are not many CEUS cases of verified adenomas reported. We have seen two verified adenomas over the years, and they followed the reported pattern of a quick, homogenous enhancement with no washout.
Adenoma Biopsy verified adenoma. MI goes from low to high and back in this exam. A bright ring of vessels can be seen in the periphery, and finally there is a late phase with no washout.
Courtesy of Dr. Knut Brabrand, Rikshospitalet University Hospital, Oslo, Norway

Detection

Metastases Up

Washout is the significant trait of metastases. The onset of visible washout may vary, but predominantly takes place at some time in the portal phase.

Washout is the basis for detection. The greater the washout, the greater the detectability. It is not unusual for conspicuous washout to take place quite late, near the late phase. Regarding very small metastases of about five mm, or even less, there must be enough time between injection and scanning to let the washout make the metastases hypoechoic enough to be detected.

90-120 seconds after injection is an appropriate starting point for the beginning of the detection Sonoexam. If scanning begins within the first minute there is a substantial risk that some metastases are not yet dark enough to be clearly seen. By waiting at least 90 seconds before scanning begins, the risk for encountering metastases that are not yet dark enough is greatly diminished.

Prescan interval monitoring of some metastases while waiting for the 90 seconds after injection may give a hint as to how fast the washout is in the actual case, and how dark metastases to anticipate in the Sonoexam.

Very slowly darkening metasteses may be a reason to return to the beginning of the Sonoexam for a second exam of the first points of the protocol.

Lack of reliable washout is very rare indeed, but has been encountered as stated under "Characterization/Metastasis" above.

General pitfalls

Haemangiomas are excellent for demonstration of technical pitfalls thanks to their predictable enhancement pattern, which initially includes a completely non-enhancing centre. There are several pitfalls examplified by haemangioma videos in this section. However, these examples serve as a reminder to consider the possibility of technical pitfalls in all CEUS procedures.

Mechanical Index mistakes Up

MI (the Mechanical Index) should be regarded as a dynamic parameter just like image gain, and needs attention throughout any exam. MI is a parameter that is new to US examiners beginning using CEUS in clinical practice. There are legal restrictions as to how high the MI may go in conventional US, but in practice the examiner never really thinks about it (except regarding the eyes which have much lower legal limits of their own). However, CEUS is a different story, and introduces MI as a very important player.

Proper MI is the responsibility of the examiner in each individual case. MI settings in CEUS using second generation microbubbles can very easily be set way too high and ruin the exam. Unfortunately this fact has to date not been emphasized enough by US machine vendors nor their application specialists. In most cases MI is set to a default value in the machines contrast programs at installation, and the users unfortunately get the impression that this value is the one to be used in future.

For lesions that are not very superficial or deep, most default MI settings will work well most of the time, but such a fixed MI may have unexpected and bad effects on the interpretations of CEUS exams if one is not aware of the problem. The effects of MI to lesions depend on many things in our experience, such as the distance from the transducer, degree of steatosis, examination time, lesion circulation speed etc.

Depth setting affects MI; increasing the image depth decreases MI.

Slightly too high an MI is the most difficult to recognize, since the negative effects may gradually but substantially affect CEUS results as microbubbles are slowly destroyed. Excessive MI has the greatest impact on superficial slowly filling haemangiomas and on the detection of superficial metastases. []

Too low an MI is usually more straightforward to recognize and is negative for CEUS of any deep lesions. It makes the image gradually dark with increasing depth in the affected areas, with no dependable CEUS echoes and disturbing noise levels by gain compensation. Of course this is negative for the detection of metastases, which depends on the contrast resolution, as well as to characterization.

"One finger always on the MI knob" is the way one ought to think about it. In 2002 when we started practicing CEUS, the MI levels were set by default to a level that permitted good penetration and gave a bright and clear signal in most cases, but we soon found that this MI level was far too high for superficial lesions and for repeated scanning in detection of metastases, with excessive microbubble destruction. Nowadays it seems that the default settings are more careful after lots of feedback to the industry.

It is impossible to compare MI settings between different machines. Some vendors display the mean MI value, while others use the peak value in their settings. An MI of 0.18 in one machine type may correspond to 0.06 in another (by experience). Another problem is that some machines change the MI level by too large increments when one turns the knob; another is that the lowest possible MI may not be low enough for very superficial lesions in thin patients.

Image noise Up

Image noise is a result of setting gain too high. As a general rule, the pre-contrast gain level should be set immediately below the level where noise is just barely seen. [noise]

Gain compensation for low MI is a mistake that is fairly common when trying to brighten an image caused by too low MI. In effect it is the image noise that increases, which should of course be avoided.
MI too low causing image noise and too bright superficial areas due to compensation by raising gain.

Too persistent scanning.Up

Scanning a lesion for a long time to see how characterization evolves after the injection of UCA may be tempting, but prolonged ultrasound exposure to the UCA has an effect on its longevity, and thereby on the clarity of the results. This is especially true after the first minute following TCA, since the inflow concentration of UCA rapidly decreases.

"Fanning" and increasing intervals is the proper Sonoexam technique to preserve the UCA optimally. “Fanning” is done over the lesion during the arterial phase, later followed by single scans through the lesion at increasing intervals, each one being recorded for later scrutiny. It is not necessary to “keep looking” bedside after the arterial phase, since the rest of the exam almost always involves the documentation of whether there is washout or not. This is more confidently done by repeated short scans through the lesion at 30 seconds intervals than by continuous scanning, since the latter does affect the microbubble concentration in the surrounding liver in a negative way and thereby rendering the washout less conspicuous. See the characterization Sonoexam protocol for suggestions.

Workstation review of the exam very often has the clue to the diagnosis if the scanning follows basic Sonodynamics principles. Remember that there should always be the possibility to look at the Sonoscans after the exam for final diagnosis.

The amount of data to be stored in PACS is greatly reduced by interval short one-direction Sonoscans of the lesion in the portal and late phases. [Too persistent scanning]

Scanning too deep Up

FLL characterization deeper than 12 cm is rarely a great success. The smaller the lesion, the worse the depth problem. There is a limit for how deep into the tissues CEUS is possible to reach, since deep lesions involve high MI settings and low spatial resolution. A particularly difficult situation is trying to characterize a small lesion close to the lung at the top of the right liver lobe, maybe because of the added effect of reflection of sound against the surface of the base of the lung.

Find a position nearer the FLL. Very often it is possible to see at least part of the lesion intercostally from a much closer position.
Characterization failure by depth Light steatosis. At long distance there is no enhancement. Moving to a closer position in late phase shows no enhancement; the slow microbubbles in the haemangioma were probably destroyed by high MI used at long distance, with added effect of reflection from lung surface. New low MI CEUS from the closer position characterizes the haemangioma.

For detection of metastases the problem is not quite as bad, since it is usually possible to detect a dark lesion against the surrounding bright parenchyma, but beyond 14-15 cm the results are less encouraging for smaller metastases.

Incorrect focal points Up

CEUS is more influenced by incorrect focus positioning than conventional US, and again MI plays a role. The sound beam is most concentrated at the levels of the focal points, which of course has an effect on the longevity of the microbubbles at that level. The deeper the focal points, the smaller the problem.

Image quality beyond the deepest focal point is more adversely affected in CEUS than in conventional US in our experience. Again, it is important to keep the deepest of the focal points as deep as possible at all times during CEUS.

Earliest phase not seen Up

TCA and the following seconds are sometimes crucial to characterization. Sometimes the very first seconds of the arterial phase may reveal characteristics of an FNH or an HCC, a very clear rim enhancement of a metastasis or a very quickly enhancing haemangioma. It is not good enough to start recording the first Sonoscan just after TCA. Sometimes a replenishment study following a high MI burst may save the situation, but one can not count on it since the replenishment tends to enhance the lesion and its surroundings and “soak” the area with microbubbles extremely quickly. If the arterial phase has been missed and the diagnosis is not obvious without it, it is safest to make a second exam when the UCA concentration has fallen.

Reflection or "glare" Up

False appearance of enhancement in a lesion is a problem whenever the absence of enhancement in a lesion is required for a conclusion. Such lesions are haemangiomas, abscesses, RF-ablations, infarctions, portal venous thrombosis and others, where a false impression of enhancement may lead to the false conclusion that there is viable tumour tissue, when in fact the absence of enhancement proves the opposite.

This problem most probably occurs due to reflection of the intense scattered contrast echo from the surrounding liver parenchyma in the lesions tissues, with added effect if the lesion is close to the reflecting lung surface or the intensely enhancing kidney.

In most cases of CEUS the problem is of minor importance, but it is important to keep the possibility of reflection artefacts in mind in order not to judge affected lesions as enhancing by mistake.

The appearance of the artefact differs from true enhancement by the fact that no "live motion" of microbubbles is seen. When the lesion is close enough for this discrepancy to be obvious, the distinction is quite clear for the experienced examiner.

Reflection artefacts increase with depth in our experience. Lesions close to the lung surface and more than 10 cm from the transducer are notoriously difficult to prove non-enhancing. Characterizations under those conditions may be impossible.

A significantly smaller dose of contrast may be an efficient remedy to the problem.
Strong glare artefact
in arterial phase of haemangioma.

Low frame rate Up

15 frames per second is the lowest limit to aim at for the arterial phase of characterization and for detection. It is important to keep the frame rate up in order to get a feeling for the “fluent” conditions of the arterial enhancement, and in order not to overlook small lesions when detecting metastases. The latter applies to both baseline and to CEUS. We have seen examples of detection exams where the frame rate has been as low as 6 fps, which is far too low and obviously causes big spatial gaps between the frames of a normal Sonoscan, where small metastases may hide.

Sacrifice a little spatial resolution to enable a higher frame rate. The great contrast resolution of CEUS compensates for this in most cases.

In extreme cases up to 50 fps have been captured to catch the very first seconds of a focal lesion. In the Sequoia this can be accomplished by selecting a ROI (Region of Interest) for the Sonoloop.