General liver CEUS technique

CEUS technique in officially recognized indications, based on our experience.

Some general considerations

Characterization of FLL's

Detection of metastases

Detection of HCC's

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

Some general considerations

CEUS, the timer and TCA.

The timer is essential to CEUS. If a timer is not used, it is practically impossible to keep track of the development of any CEUS exam. Typical metastases, for instance, show washout within a minute and microbubble degradation usually has quality consequences after 4-5 minutes. Since we prefer injecting the ultrasound contrast agent (UCA) ourselves when we examine the patients, we always make the injection 5 seconds into the timing to have time to move the left hand from the syringe to the US machine. However, for characterization of FLL’s (below) we need the arrival of the very first microbubbles into the liver image (TCA) to be the exact reference. Therefore the TCA is referred to as zero seconds in the FLL characterization chapter below, although we start the timer 5 seconds before the injection.

Mechanical Index (MI) is a critical parameter in CEUS.

Second generation microbubbles burst immediately when exposed to conventional US MI levels. However, with proper low MI settings they endure repeated scanning, but any scanning has some effect on microbubble longevity. In the average adult the choice of MI settings is pretty straightforward, with an initial MI of ~0,13 in our ultrasound machines (Siemens Acuson Sequoia™). However, one should regard MI as a variable throughout exams and be prepared to alter it if necessary. In some cases of thin or obese patients, or liver lesions in a deep or very superficial location, the MI must be adjusted for correct interpretations of the exams. Liver lesions with very slow circulation, particularly haemangiomas, are sometimes subject to faster bubble destruction than the surrounding liver parenchyma, which may give an impression of “false washout” if the MI is not adjusted to a very low level. Also, detection of washed-out metastases becomes less efficient if too high an MI is applied, which partially destroys the microbubbles in the contrasting liver parenchyma, and deeper metastases can not be seen if MI is too low to reach all the way to the distant surface of the liver. Repeated scans with increasing MI are sometimes necessary for detection of metastases in order to see both superficial and deeper parts of the liver optimally. Please read about MI in the sections below and in "Common mistakes" in CEUS peculiarities for further elaboration on MI settings.
Microbubble destruction
is always present to some degree at exposure of the UCA to ultrasound. This sequence shows destruction even at lowest MI in the arterial phase before microbubble concentration builds up. In practice this is usually no big deal unless there is long ultrasound exposure to the same spot.


Characterization of FLL’s

Characterizations begin with fundamental baseline Sonoscans of the lesion.

Scanning of a lesion without UCA is called a “baseline scan”. A baseline exam of the lesion prior to CEUS is recommended, so that the CEUS findings can be interpreted with reference to findings on baseline US.

Frame rate in arterial phase should preferably be 15 fps or more.

ln characterization the arterial enhancement pattern is sometimes quite fast, and the details become less obvious if frame rate is too low. 15 fps or above is preferred, but there are cases when up to 50 fps within a region of interest have made a difference in for example very small FNH:s, since such a technique sometimes allows for the microbubbles to be followed as to their direction of flow also in small vessels.

The lesion should preferably be less than 10 cm from the transducer.

A longer distance makes characterization less accurate. Analogous to conventional ultrasound, the closer the lesion to the transducer, the more probable it is that details of the arterial phase can be more accurately seen. It is generally better to try and reach a superficial lesion in the right liver lobe using an intercostal approach, than to try characterization from a long distance subcostally.

Take care to use a proper MI setting.

An initial MI setting of ~0,13 (Sequoia) is usually a good compromise for balance between significant bubble destruction and good penetration. However, in some cases it has to be adjusted. For very shallow lesions the MI is set as low as possible (~0,07 on Sequoia), since bubble destruction can affect the interpretation of the exam. The necessary MI increases with depth, and the MI has to be high enough so that it is not compensated for by increasing gain, which causes noise artefacts and decreases image contrast. An MI of more than ~0,18 (Sequoia with CPS) is rarely beneficial, instead a closer approach to the lesion is preferable.

Normal breathing if possible.

It is preferable that the patient can breathe normally during the exam. It is more difficult to “aim” at the lesion for the arterial phase, and to follow it through the vascular phases, if the patient has to hold his breath.

Initial positioning of the transducer to align the centre of the lesion with any obvious feeding artery.

Many FNH's and some HCC's have conspicuous arterial enhancement patterns. Both sometimes have an identifiable main feeding artery which preferably is positioned at the very periphery of the lesion in the initial field of view by turning the transducer to the proper angle with the aid of colour Doppler before the CEUS. In this way chances are that the further course of the artery can be detected in the arterial phase of CEUS.The scanning pattern during the very first seconds of the arterial phase is important to understand the local circulation pattern. With experience, the examiner can often adjust the initial scanning with respect to the findings during the first seconds to achieve the most information. However, basically the transducer should be placed covering the lesion, and if possible along any obvious feeding artery at TCA. A few seconds following the arrival of contrast to the region of interest it is usually quite evident whether the lesion is probably avascular, a slowly filling haemangioma or something else that is vascular and thus enhances in the very first seconds. It should also be assessed if the MI setting seems adequate. If there is very good signal from the UCA beyond the lesion then a lower MI should be considered in order to preserve microbubbles for later in the exam. On the other hand, if there is no echo from the UCA in the liver surrounding the lesion the MI is too low (see the Mechanical Index section under pitfalls).

"Fanning" during the initial enhancement of characterization, then short scans at longer and longer intervals.

For characterization of FLL it is important to try and see the very first seconds of the arterial enhancement of the lesion in detail, but the scan plane can only be at one place at a time. This limitation has to be overcome by the examiner at the patient’s bedside by scanning back and forth during the arterial phase through the lesion with the scan plane. Let us call it “fanning” for simplicity. The arterial phase of CEUS is an exception from the Sonodynamics' rule of scanning in one direction only. Generally fanning back and forth begins 1-3 seconds following TCA. The speed of the fanning is roughly 2 seconds to cover a 3 cm big nodule and 3 seconds over a 5 cm nodule. As always, assessment of the exam at a workstation or the machines monitor should be possible. Try to make the first fanning motion out of the lesion in the cranial direction or to the left, so that the following complete fanning through the lesion goes craniocaudally or left-right. Such structure makes comparisons between different exams more straighforward.

First scrutiny of the arterial phase, then the portal and late phases.

Ultrasound's time resolution is a very good asset, but for high frame rates the CEUS exams are limited to one scan plane with current technique. The initial enhancement is important for confident characterization of all kinds of FLL's. Unfortunately not all FLL’s display “typical” behaviour, but optimization of exam technique may sometimes be the clue to dependable characterization in those that do. Because of the very quick series of events that occur in the arterial phase, it is important to start the acquisition of the first Sonoscan at TCA. As the exam proceeds beyond the arterial phase, constant monitoring is no longer necessary, and Sonoscans can be made at longer and longer intervals. This way microbubble destruction is minimized, and the exams occupy less storage space. Make careful preparations regarding the frame rate, transducer position, region of interest, focal depth, gain, Mechanical Index (MI) and consistent Sonoscans. It is important to be able to re-evaluate the exam at a good workstation or the monitor of the US machine. Unfortunately there are HCC's which do not wash out and old FNH's and some haemangiomas that do wash out. These anomalies are reasons to optimize the arterial phase parameters as far as possible.

Bubble bursts with replenishment studies can provide "additional initial enhancements" in small volumes.

Sometimes it works very well to repeat the initial enhancement of a fast filling lesion by one or a few high MI burst, that momentarily disrupt all microbubbles. Immediately after the bursts the initial enhancement pattern of the lesion is revealed by the new microbubbles entering the scene. The most efficient way to create bursts is by using short colour doppler exposures to the region of interest. This technique is not always beneficial, but it can at times be the key to correct characterization of rapidly enhancing lesions or in cases when the transducer was not exactly on target at TCA. Again, high framerate and second-look capability of the Sonoexam are a must.
Bubble burst and high MI "pseudoangiogram"
of small high flow haemangioma, with slowmotion workup of bubble burst (see following point).
Courtesy of Dr. Anna-Karin Siösteen-Tofte, Karolinska University Hospital, Sweden.
High MI and bubble burst replenishment in small haemangioma. Both techniques reveal the arterial characteristics of a haemangioma.

High MI angiograms may help understanding small arterial trees and direction of enhancement.

By examining a fast, hypervascular lesion with a very high MI, the microbubbles burst as they enter the field of view. In appropriate circumstances this technique makes the arteries of the lesion stand out against a background of less enhanced tissue, thus creating a type of "pseudo angiogram". This has been helpful in cases of very quickly enhancing lesions where the vessels otherwise quickly "drown" in the tissue perfusion, giving some more time to study the arteries, mainly in fast haemangiomas and FNH's. However, it is not easy to predict which lesions will benefit. Sometimes a reinjection with a very small amount of UCA (0,3 - 0,6 ml) and an initial MI of about 0,6 to 1,0 makes the process easier to control. Furthermore, the technique is more beneficial for lesions close to the transducer than lesions at greater depths.
High MI angiogram in FNH
High MI used initially to emphasize the arterial tree of the lesion. "Spoke wheel" arterial tree, and at low MI enhancement from the vessel areas as well as a central scar. No washout.


Detection of metastases

Frame rate in detection should preferably be 15 fps or more.

When scanning the liver with normal Sonoexam speed it is important not to use too low a frame rate, since we have found that thick "slice increments" between images may lead to jumping over small metastases in the Sonoloops.

Detection of metastases with two liver Sonoexams: One US, and one late phase CEUS after 90-120 seconds following injection.

The goal is to find washed-out metastases, and the earlier CEUS phases usually give very little additional information and are not routinely scanned. The first US Sonoexam sometimes gives valuable additional information, for example of simple cysts, obvious metastases, or very bright or calcified metastases that may not be conspicuous on CEUS. The initial Mechanical Index of the Siemens Acuson Sequoia™ is about 0.13, but this setting may have to be altered when scanning the right liver lobe in order to penetrate all the way to the diaphragm. If in doubt whether the "survival time" of the microbubbles (usually 4-6 minutes) has been exceeded at some point of the protocol, it is essential to make another injection of 2,4 ml of UCA and wait for 90 seconds before proceeding with the rest of the protocol. A second full dose injection can be made three to four minutes after the first one, and scanning can continue after another 90 seconds. A third injection has never been necessary unless some equivocal finding required characterization after detection. If this is the case, it is advisable to wait about ten minutes before the third injection to give the UCA time to vanish first. When using the Siemens Acuson Sequoia™, we currently prefer using the CCI mode rather than the CPS mode for detection of metastases, based on our experience of their respective frame rates, spatial resolutions, contrast resolutions and impact on microbubble longevity. For some unknown reason the UCA lasts longer in the liver in some patients than in other. Also, the times at which the portal and liver veins become hypoechoic also differ greatly among individuals. Some livers maintain bright veins for the duration of the exams, while others show darkening of the veins after only 2-3 minutes, still leaving the parenchyma quite brightly enhanced. However, thanks to the tubular shape of the veins it is usually not difficult to distinguish metastases from veins, although detection is easier in cases with isoechoic veins.
Why wait 90 seconds?
Demonstration of washout in metastases over time.

Repeated Sonoscans with different MI is necessary in areas where the initial MI setting is not sufficient.

As an average, an MI of about 0,13 (Sequoia) is sufficient for examination through the depth of the left liver lobe, including segment 1, without significantly destroying superficial microbubbles. However, depending on the size of the patient and the general attenuation of the, the initial MI may have to be adjusted up or down in some cases. Of course, too low an MIBegin with an MI that penetrates as deeply as possible without destroying superficial microbubbles. When reaching deep liver areas, typically the right liver lobe covered by Sonoscans 6 and 7, the initial MI setting is often too low to penetrate all the way to the diaphragm, resulting in insufficient brightness in the deep parts of the liver for sensitive perception of dark metastases. When this happens we increase the MI as we proceed at MI increments of roughly 0,07 (Sequoia), and eventually return to a Sonoscan for a repeated scanning in order to see through the depth of the liver. In the repeated scans we accept that there is bubble destruction in the superficial parts of the liver, since these have been previously covered with a lower MI.
Detection with increasing MI At the end an example of bubble destruction due to going from higher to lower M
MI decrease with depth increase MI decreases when going from superficial to deep field of view.

Further reading:

For further details on our technique for detection of metastases, please see


Detection of HCC's

MI setting far above usual in arterial phase.

Compared to metastases, HCC's do not have a reliable washout, and some may not wash out at all. On the other hand they are practically always hypervascular, and enhance earlier than the liver parenchyma. The latter is true even in cases of cirrhosis, where the liver itself generally has an earlier enhancement than normal. Regenerative nodules enhance later than the cirrhotic liver parenchyma and become isoechoic in the portal phase. Thus, the detection of HCC's and regenerative nodules has to occur in the arterial phase. HCC's may become more conspicuous if MI is raised to an otherwise "banned" level of about 0.5-0.7 (Sequoia) during this scan. This way the rapid enhancement of HCC's often stands out more clearly against the relatively less enhancing liver parenchyma. One must remember to decrease MI to a regular CEUS MI level after this initial arterial phase scan. Following eventual microbubble replenishment studies, a conventional liver Sonoexam is performed starting after 2,5 minutes (washout in HCC's is usually late).