Imaging of Liver Tumors

Imaging of Liver Tumors

Michael L. Wells, MD

Sudhakar K. Venkatesh, MD

Radiologic evaluation represents a critical component of liver tumor detection and characterization. After discovery of a liver mass, the radiologic findings will typically provide guidance for patient management. Management may include imaging follow-up, percutaneous biopsy, or surgical resection. Radiologic findings can safely diagnose benign lesions such as cysts or hemangiomas with sufficient confidence that further workup is unnecessary. However, for most complex cystic and solid liver masses the imaging findings are often not entirely specific, and a differential diagnosis of pathologic entities are provided. When a benign entity is favored, many patients will be given imaging follow-up to determine stability of the mass over time. In challenging cases, a mass discovered by imaging must ultimately be diagnosed by percutaneous biopsy or surgical resection. Radiologic-pathologic collaboration, with careful correlation between the initial imaging findings and the pathology findings, is often helpful to arrive at the correct diagnosis. Radiologic imaging is very helpful for describing the size, anatomic extent, and number of masses, which may not be entirely included within a biopsy or surgical specimen. Correlation with prior imaging is often necessary to ensure diagnostic concordance and can prevent errors in diagnosis by revealing inadequate sampling of a tumor or sampling of the wrong lesion.

The vast majority of liver mass evaluations entail cross-sectional imaging using ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI). A basic understanding of the principles of operation and the terminology used to describe the images is helpful for each of these modalities. A brief introduction of these modalities is provided below. Liver mass evaluation is further supplemented with additional radiologic modalities including positron emission tomography (PET), scintigraphy using a variety of radiopharmaceuticals, and fluoroscopic angiography. Discussion of these modalities will be limited and provided in the context of individual pathologic liver entities.



Ultrasound (US) is often the first-line test used to examine the liver because of its low cost, lack of ionizing radiation, and accessibility. US produces images by imparting sound waves into the tissues at a high
frequency, typically between 3 and 7 MHz. Lower frequencies travel further into the tissues, allowing deeper tissues to be visualized, whereas the higher frequencies provide better spatial resolution and anatomic detail with more limited depth of penetration. As sound waves travel into the liver, they encounter tissues that transmit the sound waves at different speeds (e.g., transmission speed in liver is 1,570 m/s, in water is 1,489 m/s and within fat is 1,450 m/s). Interfaces between these tissues result in a change in transmission speed and the reflection of sound waves back to the transducer, where they are detected and used to form an image. The US images of a liver mass are described in terms of relative echogenicity, with areas of the mass that are brighter than the background liver referred to as hyperechoic, areas that are similar in brightness referred to as isoechoic, areas that are darker referred to as hypoechoic, and areas with an absence of any signal referred to as anechoic. It is important to note that the background liver can have abnormally increased echogenicity in the setting of steatosis or fibrosis, which can alter the US appearance of liver masses.

Doppler imaging is used to assess blood flow within the liver. As sound waves encounter moving elements within the blood, not only are the sound waves are reflected back to the transducer, but the movement of the object alters the frequency of the reflected waves. This phenomenon is analogous to the higher pitched sound of a train horn while it is moving toward you and lower pitch while it is moving away. Detecting this change in frequency allows calculation of the direction and velocity of the flowing blood. This property can be used to assess flowing blood within the vasculature and within liver tumors.

The strengths of US include its ability to assess the liver parenchyma, biliary tree, and vascular supply of the liver. In the evaluation of liver masses, US is primarily useful for mass detection and for its ability to differentiate between cystic and solid lesions. US is limited by its inability to penetrate bone or gas filled organs. Overlying ribs and gas-containing bowel or lung tissue occasionally results in regions of the liver that cannot be adequately visualized. The depth of sound wave penetration is also limited, which can lead to incomplete parenchymal evaluation. This is particularly a problem in obese patients, where there is increased thickness of subcutaneous fat, and in livers with poor penetration because of parenchymal disease such as steatosis or fibrosis. Of note, liver evaluation with US is highly operator dependent. Sonographers working in a low-volume practices may be prone to inadequate evaluation of the entire liver parenchyma, or not optimizing the scanner settings to demonstrate cystic components of a mass or internal blood flow.

Computed tomography

Computed tomography (CT) is commonly used to evaluate liver tumors because of its accessibility, reproducibility, and its relative simplicity to perform and interpret. CT produces images by generating high-energy photons (typically up to 140 keV), which are a transmitted through the patient and are received by a detector on the opposite side of the patient. The photon generator and detector are mounted on a gantry, which rotates around the patient in a helical pattern. Tissues of different composition attenuate photons to varying degrees. These differences in attenuation result in contrast, which is used to generate a CT image. Bright regions of a CT image are referred to as high attenuation or hyperdensity and dark regions as low attenuation or hypodensity. Attenuation of tissue substances on CT images are often measured in terms of Hounsfield units (HU). Hounsfield units are a measure of relative attenuation and calculated by the formula HU = 1,000 × (µµwater)/(µwaterµair), where µ is the linear attenuation coefficient. For reference, common HU include air = -1,000, fat = -100 to -50, water = 0, blood = 30 to 45, liver 40 to 60, and bone 700 to 3,000.

Iodine-containing contrast medium is routinely injected intravenously into the patient to increase contrast between tissues. Iodinated contrast injection provides information regarding the anatomic vascular supply, the rate of blood flow, and the volume of distribution within tissue. These characteristics are very useful for describing and categorizing liver masses. Increased attenuation of tissue in response to IV contrast uptake is referred to as enhancement. In the first few minutes following contrast injection, the relative amount of iodinated contrast changes rapidly within tissues of different composition. This change is typically assessed by scanning the patients multiple times to provide a dynamic evaluation.

The time at which a patient is scanned after injection is referred to as the phase of contrast enhancement. The most important phases of contrast enhancement for liver tumor characterization include the late arterial phase, the portal venous phase, and the delayed (or equilibrium) phase. The late arterial phase is the time at which the injected contrast bolus has opacified the hepatic arteries and has just begun to opacify the portal veins; this typically occurs 15 to 35 seconds after contrast injection. Imaging at time points earlier than this may fail to show enhancement in tumors with hepatic arterial vascular supply. The portal venous phase is defined as the time at which the contrast bolus has opacified the portal veins and has just begun to opacify the hepatic veins; this occurs roughly 60 seconds after injection. This is typically the time at which the hepatic parenchyma is maximally
enhanced. The delayed or equilibrium phase includes imagings that are usually performed between 3 and 10 minutes after injection. Most iodinated IV contrast agents used for CT remain within the extracellular compartment of tissues. Over time, these contrast agents diffuse through the extracellular space to equilibrate between the extracellular intravascular and extravascular compartments. Fibrotic or sclerotic tissues remain enhanced in the delayed phases because of large extracellular spaces associated with fibrosis.

CT has several advantages, including its ability to scan rapidly and to provide a comprehensive assessment for metastatic disease. CT examinations are often performed for liver mass evaluation out of convenience, as the entire body can be quickly assessed for metastatic disease in a single setting. The ability to acquire an imaging data set within a few seconds is a major advantage for patients who cannot hold their breath or cannot breathe consistently, as is required for MRI. An inability to adequately suspend respiration can render an abdominal MRI nondiagnostic. CT is also able to acquire high spatial resolution, 3D image data sets that can be reformatted in any plane. These types of data sets are difficult to acquire at MRI and, with current technology, most MRI images are acquired with the intention of being viewed in a single plane.

Specific limitations of CT include inferior soft-tissue and IV contrast resolution when compared with MRI. The limited soft-tissue contrast can create difficulty differentiating a low-attenuation solid mass from a cystic mass or lipid-containing tissue. Similarly, high-attenuation tissue or calcification can occasionally be difficult to differentiate from iodine-containing contrast media. Despite these limitations, CT has a very high sensitivity for detection of small calcifications and gas, when compared with either MRI or US. The radiation exposure imposed by CT is another limitation, one that frequently restricts the number of phases obtained during an examination and the frequency of follow-up exams. However, in the presence of malignancy or in advanced patient age, the risk imposed to the patient because of radiation exposure at CT is probably negligible.

Magnetic resonance imaging

MRI provides the best discrimination between soft tissues of different types and is the modality of choice for both liver mass detection and liver mass characterization. MRI uses a strong static magnetic field, weak gradient magnetic fields, and radiofrequency transmission coils to alter the orientation of protons found within hydrogen atoms, which are the most abundant element in the human body. Protons are imaged by specifically changing gradient field strengths, applying energy through radiofrequency transmissions, and recording signal from the tissue at varying time points. The combinations of the steps required to produce images are referred to as pulse sequences.

In a strong static magnetic field, the axis of a proton’s magnetic dipole will tend to align with that of the static magnetic field. The axis of a proton’s dipole is not aligned exactly with that of the magnet but slightly off center and precesses (rotates) around it at a frequency proportional to the strength of the magnetic field. This precessional frequency is specific to particular protons and can be exploited by using an applied radiofrequency electromagnetic pulse at the same frequency to shift proton orientation away from that of the magnetic field. After protons have been shifted, they will relax (or reorient) with the direction of the static magnetic field.

There are two basic ways in which protons relax after being stimulated by a radiofrequency pulse: T1 (or longitudinal) relaxation and T2 (or transverse) relaxation. These two types of relaxation occur at different rates, which are affected by a proton’s surrounding magnetic micro-environment or tissue type. Measuring differences in the T1 and T2 relaxation rates between different tissues creates contrast. This contrast in turn forms the basis for creating images at MRI. When reviewing MRI images, bright regions of the image are referred to as having high signal or hyperintense and dark regions as low signal or hypointense.

A typical liver imaging protocol includes a combination of pulse sequences designed to show contrast differences between T1 and T2 relaxation rates of different tissue and are referred to as T1-weighted and T2-weighted images. Images of each sequence are specifically used to reveal different anatomic features or abnormalities of tissue content, so characterization of a tumor using MRI requires review of several image sets.

A basic knowledge of typical tissue signal intensities on MRI images is necessary for image review. On T1-weighted images, water is low in signal and fat is high in signal. On T2-weighted images, both water and fat are high in signal. The weighting of an image can often be determined by looking at the signal of cerebrospinal fluid in the spinal canal. T2-weighted images are predominantly used to image water content of tissues. An extreme example of T2-weighted image is magnetic resonance cholangiography, in which the images are very heavily T2-weighted, such that the only structures clearly visible on the images are fluid-containing structures, such as the bile ducts and renal collecting system. It is important to know that both T1- and T2-weighted images are often altered by suppressing the signal from fat. This is accomplished by a variety of techniques and allows
increased dynamic range of the images, increased conspicuity of liver parenchyma, and detection of fat-containing structures.

Specialized MRI techniques that are used to image the liver include in-phase and opposed-phase imaging and diffusion-weighted imaging. Protons within water precess at a slightly different frequency than those found in fat. When T1-weighted image acquisition is timed so that the magnetic dipoles of both water and fat are oriented in the same direction, the signal intensities from both tissue types are additive; this produces an in-phase image. When the timing of image acquisition is altered so that the dipoles of fat and water are oriented in opposite directions, the signal intensities cancel out and the overall signal intensity decreases, this produces an opposed-phase image.

These properties can be exploited to identify tissues that contain elements of both water and fat. When the signal from a voxel (a position in a 3 dimensional image) of an image of tissue originates from 50% water and 50% fat, the signal intensity of that voxel should be near zero on the out-of-phase image. Because of this, lesions that drop in signal intensity on the opposed-phase image, when compared to in-phase image, are likely to have fat.

Diffusion-weighted imaging basically measures the Brownian motion of water molecules. Tissues that are either very cellular or swollen tend to have restricted motion on diffusion-weighted imaging. Diffusion-weighted imaging is a technique in which specialized gradient magnetic fields are added to a T2-weighted image. These gradients fields are designed such that bound protons that are unable to move (their ability to diffuse is restricted) do not lose signal, whereas moving protons lose signal rapidly. The most common explanation for restricted diffusion in tissues is increased cellularity, in which a greater percentage of the imaged protons are within cells and their movement is physically restricted by cellular membranes. Another common reason for restricted diffusion includes abscess formation, in which the protons are extracellular but free diffusion is restricted by the highly viscous and proteinaceous environment. The resulting diffusion-weighted image shows areas of high signal, which could represent either an intrinsically T2 hyperintense tissue or a tissue with restricted diffusion. Multiple diffusion-weighted images are produced with increasing strengths of the gradient fields. The relative change in signal of each pixel over the series of images is used to calculate an apparent diffusion coefficient (ADC) map. The apparent diffusion coefficient map shows pixels on the diffusion-weighted image, with true restricted diffusion showing low-signal intensity, whereas regions of high-signal intensity that are seen on both diffusion-weighted image and the ADC map represent tissue with intrinsically high T2 signal (on the diffusion-weighted image these regions are referred to as T2 shine through) and/or unrestricted diffusion.

IV contrast-enhanced MRI imaging of the liver is typically performed with fat saturated T1-weighted sequences. IV gadolinium shortens the T1 relaxation time of tissues, resulting in higher signal intensity. Most gadolinium contrast agents are extracellular and have dynamic enhancement properties similar to those discussed previously for iodinated CT contrast agents. The phases of dynamic contrast enhancement that are routinely acquired at MRI are the same as those acquired at CT. MRI contrast agents, referred to as hepatobiliary agents, are also available. These agents are actively taken up into hepatocytes and excreted into the bile ducts. The most commonly used is disodium gadoxetate (Gd-EOB-DTPA; Eovist; Bayer Corporation, Pittsburgh, PA) and gadobenate dimeglumine (Gd-BOPTA; MultiHance; Bracco Diagnostics Inc, Princeton, NJ). These contrast agents are typically imaged in a dynamic fashion similar to extracellular agents. Additional images at delays of 20 to 120 minutes are referred to as the hepatobiliary phase. During the hepatobiliary phase, the normal liver is diffusely high signal and contrast agent increases the signal intensity of the biliary tree. These images have a high sensitivity for detection of liver masses that do not contain hepatocytes and thus do not retain the contrast agent.

MRI is considered the preferred modality for liver mass evaluation.1 MRI pulse sequences are exquisitely sensitive to changes in normal liver parenchymal composition and are highly sensitive for detection of liver masses.2, 3, 4 MRI is more sensitive than CT or US for detection of substances such as fat or iron.3 Despite these advantages MRI does have limitations, including cost and patient contraindications to the MRI scanner. Many metallic implanted foreign bodies and pacemakers are not compatible with the strong magnetic fields and radiofrequency gradient pulses. Patient inability to cooperate can also result in a severely limited MRI exam. Acquisition of MRI images takes a relatively long time and patients who cannot hold their breath or cannot breathe at a consistent rate will result in images with substantial motion artifact. MRI is also limited by a number of image artifacts because of magnetic susceptibility of metallic foreign bodies, pulsation artifacts from vessels, and other artifacts when patients are large in size or have significant ascites. A detailed discussion of these artifacts is beyond the scope of this chapter, but it is important to recognize that these artifacts can impair the interpretation of images and occasionally simulate liver pathology.


Simple hepatic cyst

Simple liver cysts are readily diagnosed on imaging. By US evaluation, the contents of a simple cyst should be anechoic (Fig. 24.1). The cyst walls should be thin or imperceptible. The lack of internal architecture to attenuate sound waves results in higher amplitude of echoes in the tissues directly posterior to the cyst, a phenomenon referred to as posterior acoustic enhancement. There should be no detectable Doppler vascular flow within a cyst. CT findings of a homogenously low-attenuation mass (0 to 20 HU) with sharp borders and no internal enhancement are consistent with a benign cyst.5 MRI will show a cyst to be homogenously high signal on T2-weighted images; low signal on T1-weighted images with no detectable internal enhancement. A simple cyst will be high signal on both diffusion-weighted imaging and apparent diffusion coefficient, reflecting its “T2 shine through” effect and the lack of true restricted diffusion. At hepatobiliary phase imaging, excretion of contrast material into the biliary tree will not demonstrate uptake into a simple cyst, because of lack of biliary communication.

Figure 24.1 (A) Simple cyst demonstrated on US. Typical features include an anechoic cyst (asterisk) with thin or imperceptible walls (arrowheads) and posterior acoustic enhancement (arrows). (B) Grayscale US performed on the same patient 6 years later shows increased size of the mass, but similar imaging findings diagnostic of a benign cyst. (C) Axial image from a contrast-enhanced CT scan performed to evaluate the cyst shows CT features of a simple cyst (asterisk) including homogenous internal fluid (0-20 HU) attenuation, thin or imperceptible walls, and lack of enhancement. (D) Coronal T2-weighted MRI image showing a simple cyst (arrow) with very high T2 signal identical to that found in the gallbladder (arrowhead) and bowel lumen (dashed arrow). (E) Axial T2-weighted MRI image shows the same cyst (arrow) to be much higher in signal intensity when compared with a nearby adenocarcinoma metastasis (dashed arrow). (F) Axial fat-suppressed T1-weighted image shows the cyst (arrow) to be very low in signal intensity consistent with simple fluid. The cyst is much lower in signal when compared to the nearby metastasis (dashed arrow). (G) Axial fat-suppressed T1-weighted image with intravenous contrast enhancement in the portal venous phase shows no enhancement of the cyst (arrow), whereas the metastasis (dashed arrow) shows irregular peripheral increased signal because of accumulation of contrast.

The finding of a simple cyst in a subcapsular location of hepatic segment IV suggests that it is a ciliated foregut cyst.6 The finding of numerous tiny (<15 mm) simple cysts within the liver is suggestive of biliary hamartomas (von Meyenburg Complexes).6
Numerous larger hepatic cysts (typically several centimeters in size) can be seen with autosomal dominant polycystic liver and kidney disease. In this condition, the intraparenchymal liver cysts can be accompanied by numerous small peribiliary cysts and cysts within the kidneys. Cysts that have hemorrhage can result in increased echogenicity at US, increased attenuation at CT, and increased T1 signal at MRI. Cysts complicated by infection can have imaging findings identical to an abscess.

Figure 24.2 (A) Grayscale US image of a hepatic abscess. Abscess cavity (asterisk) is hypoechoic surrounded by an irregular wall (arrow). A thin low-echogenicity band surrounds the abscess cavity corresponding to edematous hepatic parenchyma (dashed arrow). (B, C) Axial fat-suppressed T1-weighted contrast-enhanced MRI image (B), axial fat-suppressed T2-weighted MRI image (C), and axial T1-weighted MRI image (D) show similar findings including a nonenhancing central cavity (asterisk) surrounded by an irregular enhancing abscess wall (arrows) and a band of low T1 and high T2 signal edema (dashed arrows). (E, F) Axial diffusion-weighted image (D) and apparent diffusion coefficient map (E) show high and low signal of the abscess cavity (asterisk) respectively indicated marked restricted diffusion. Structures incompletely included on these images (B-F, arrowheads) represent septic thrombophlebitis within the hepatic veins.

Figure 24.2 (continued)


A pyogenic liver abscess can occasionally simulate a liver tumor at imaging. The lesions are often multiple and may be clustered in one region of the liver, becoming confluent over time. Pyogenic abscess because of Klebsiella pneumoniae are typically multiseptated or multi-loculated whereas an amebic (Entamoeba histolytica) liver abscess is unilocular and often single. At US, liver abscesses tend to have poorly defined borders because of adjacent inflammation (Fig. 24.2). The internal contents vary in appearance but they typically appear hyperechoic early and become hypoechoic over time, sometimes with the development of internal septations.7 Despite internal echogenicity, abscesses show posterior acoustic enhancement, which is a clue to its internal fluid content. CT typically demonstrates a hypoattenuating mass with an irregular margin. Early in abscess formation, the lesion may enhance homogenously. After necrosis has occurred and fluid has accumulated, adjacent inflammation and hyperemia will result in rim enhancement and central avascularity. Inflammation can cause transient segmental perfusion abnormalities in the adjacent liver, which are characterized by geographic regions of arterial hyperenhancement, which fade to isoattenuation on delayed images. CT identification of internal gas in the absence of instrumentation or air introduction from the biliary tree is diagnostic of an abscess. On MRI, an abscess is T1 hypointense and T2 hyperintense.8 Reactive edema within the adjacent hepatic parenchyma appears as a hazy T2 hyperintense halo. As with CT, rim enhancement and adjacent perfusion abnormalities may be present. An abscess typically results in significant restricted diffusion, which can be a helpful diagnostic finding.

Numerous tiny hepatic microabscesses (<2 cm) can be seen with conditions such as disseminated candidiasis and staphylococcus septicemia.9 These cases may show similar findings in the spleen and lungs. The lesions appear as small hypoechoic or hypoattenuating nodules at US and CT respectively. At MRI, they appear similar to larger pyogenic abscesses in that they are T2 hyperintense and show restricted diffusion. The imaging differential diagnosis for these findings typically includes metastatic disease, sarcoidosis, and lymphoma.

Focal hepatic steatosis and focal fatty sparing

Hepatic steatosis shows a variety of patterns on imaging. The changes are most commonly diffuse, but can also appear segmental, perivascular, or focal. Focal steatosis can simulate a mass on both US and CT imaging (Fig. 24.3). MRI using in and opposed-phase imaging is very sensitive to the presence of fat and the diagnosis is often readily apparent.10 On US, regions of hepatic steatosis are hyperechoic relative to normal parenchyma and at CT they are hypoattenuating compared to normal parenchyma. Conversely, a diffusely steatotic liver with a region of focal sparing can simulate a hypoechoic mass on US, and on CT can appear as a hyperattenuating or hyperenhancing mass. At imaging, regions of steatosis can show
focal nodular or segmental patterns that simulate a malignancy and should be considered in the differential diagnosis of focal lesions. Clues to the correct diagnosis include a lack of mass effect on adjacent liver borders, undisturbed course of bile ducts and/or blood vessels, geographic and/or linear borders, and typical location along the gallbladder fossa and falciform ligament. When biopsy of a suspected mass reveals normal liver parenchyma, careful review of the patient’s imaging should be performed to exclude the possibility of a benign condition, such as focal steatosis, and to ensure the biopsy was performed at the proper location.

Figure 24.3 (A, B) Apparent hyperechoic mass in the right hepatic lobe discovered on US (A, arrows). The mass was further evaluated at CT and shown to be focal hepatic steatosis (B, arrows). Characteristic findings include geographic pattern of distribution, angulated margins (A, dashed arrow), and lack of mass effect with portal veins (B, arrow heads) coursing normally through the parenchyma. (C-E) Axial hepatobiliary agent contrast-enhanced MRI obtained at 20-minute delay shows an apparent mass within hepatic segment IV (C, arrow). The abnormality is barely visible on axial T1-weighted in-phase image (D, arrow). Significant loss of signal on axial T1-weighted out-of-phase image (E, arrow) shows the lesion to be composed of intracellular lipid. Not shown, the absence of a fat-containing mass was further proven in this case by lack of concerning findings on T2 and diffusion-weighted sequences.

Cavernous hemangioma

Cavernous hemangiomas typically appear as well circumscribed, round or lobulated mass. On US, hemangiomas are typically homogenously hyperechoic with posterior acoustic enhancement (Fig. 24.4). The
diffuse hyperechogenicity is thought to be caused by reflections from the walls of numerous internal blood filled channels. In 20% to 40% of hemangiomas, US shows a peripheral rim of hyperechogenicity and central hypoechogenicity, a less common but recognized pattern.11 At CT, hemangiomas are typically homogenously hypoattenuating on precontrast images. Dynamic contrast enhancement on CT and MRI shows a characteristic pattern of peripheral discontinuous globular enhancement with progressive centripetal enhancement (moving toward the center) on subsequent phases.5 Hemangiomas tend to fill in completely and retain contrast on delayed-phase images. On all phases, the appearance of the contrast within the hemangioma is similar to that of the aortic blood pool.12 This characteristic enhancement pattern is considered highly specific for hemangiomas.13 Small hemangiomas and atypical larger hemangiomas may “flash fill,” in which the entire lesion enhances homogenously, but again the degree of the enhancement is similar to that of the aorta. MRI is very helpful in the diagnosis of hemangiomas by demonstrating high T2 signal, near that of simple fluid. Combining the findings of high T2 signal with characteristic enhancement pattern allows a confident diagnosis to be made in most cases.

Figure 24.4 Typical hemangioma. (A-C) Axial T1-weighted contrast-enhanced MRI images in the late arterial (A), portal venous (B), and 15-minute delayed phase (C) show the classic peripheral, nodular, discontinuous enhancement with progressive centripetal central enhancement. Importantly, the hemangioma is isointense to the aortic blood pool on all images, retains contrast even when imaged at a long delay, and fills completely. (D) Axial T2-weighted fat-saturated MRI image shows the characteristic high T2 signal of a hemangioma, which is similar to free water in the biliary tree (dashed arrow) and spinal canal (arrow head). (E) Grayscale US image shows the typical well-circumscribed and homogenously hyperechoic appearance of a hemangioma. Note also the finding of posterior acoustic enhancement, which is often also seen with hemangiomas (arrow head). (F) Angiogram performed by injection of the common hepatic artery shows three separate hemangiomas with peripheral nodular contrast enhancement (each hemangioma is indicated by a separate arrow).

Large hemangiomas (greater than 5 cm in size) are often referred to as giant hemangiomas. These tend to show atypical enhancement, with lack of complete central filling because of the presence of a central scar.14,15 The central portion of the lesion, which does not enhance, is typically markedly hypoattenuating on CT, markedly T2 hyperintense, and T1 hypointense at MRI (Fig. 24.5). Larger lesions can also demonstrate inhomogeneous T2 signal intensity because of regions of thrombosis or fibrosis.16 CT can occasionally demonstrate internal calcification within a giant hemangioma.14 Hyalinized or sclerosed hemangiomas do not have the typical T2 signal characteristics or enhancement patterns of conventional hemangiomas

and they are often difficult to adequately characterize on imaging.17

Figure 24.5 Atypical appearances of hemangioma. (A, B) Grayscale US image of a hepatic hemangioma showing a frequently seen atypical appearance in which the periphery of the lesion is hyperechoic (arrows) and central portion is isoechoic or hypoechoic. Corresponding portal venous phase axial CT scan image shows the characteristic peripheral globular enhancement pattern of the hemangioma (arrow, B). (C, D) Grayscale US image showing a hemangioma as a hypoechoic mass (C, arrow). The appearance is caused by a diffusely steatotic liver, which abnormally increases the echogenicity of the background parenchyma. Corresponding axial image from a CT scan performed in the portal venous phase shows the hemangioma (D, arrow) in an abnormally low attenuation liver. (E-G) Axial CT images in the precontrast (E), portal venous (F), and delayed phases (G) showing a giant hemangioma with a central scar (E, arrow) and typical peripheral progressive enhancement of a hemangioma (F, G, arrows). (H) Sulfur colloid scintigraphy of the giant hemangioma shows radiotracer uptake diffusely throughout the hemangioma (arrows) except for the central scar.

Figure 24.5 (continued)

Radionucleotide imaging with technetium labeled red blood cells (Tc-RBC) is only rarely performed to confirm the diagnosis of cavernous hemangioma. In Tc-RBC studies, the typical scintigraphic findings include poor radiotracer uptake of the hemangioma relative to the adjacent liver on immediate images, and relatively increased tracer uptake on delayed images, which may be performed several hours after tracer injection.18 In large hemangiomas, areas of poor delayed tracer uptake can be present. These areas correspond to the regions of scarring/thrombus better seen on cross-sectional imaging with CT or MRI.

Focal nodular hyperplasia

Focal nodular hyperplasia typically presents on imaging as a solitary well circumscribed mass. The US appearance of focal nodular hyperplasia is nonspecific and most commonly shows an isoechoic or slightly hypoechoic mass.19 Doppler analysis can demonstrate a feeding artery entering the center of the mass, which branches into a stellate or “spoke-wheel” pattern (Fig. 24.6). On CT and MRI, focal nodular hyperplasia appears similar to the background parenchyma, being nearly isoattenuating at CT, and typically slightly T1 hypointense to isointense or slightly T2 hyperintense to isointense.

Because of their similar appearance to the liver parenchyma, focal nodular hyperplasia in some cases can remain undetected on imaging, which led to their name “stealth lesions.” The lesions may have a characteristic central scar, reported to occur in 35% of lesions less than 3 cm and 65% of larger lesions.20 When present, the central scar is considered to be a specific finding.21 The central scar appears hypoattenuating on CT images and T2 hyperintense on MRI images. The characteristic dynamic contrast enhancement appearance of focal nodular hyperplasia at CT and MRI is a homogenously hyperenhancing mass on late arterial phase images, which fades to isointensity/isoattenuation at delayed-phase imaging.20,22,23 The central scar enhances poorly on early phase imaging but characteristically shows delayed-phase enhancement.

Occasionally, enlarged feeding arteries are associated with the lesion. Focal nodular hyperplasia occasionally contains detectable fat at MRI, although this is considered unusual, and it often occurs most often in the setting of background hepatic steatosis.24 Imaging with hepatobiliary agent is useful for making the diagnosis of focal nodular hyperplasia. This is particularly important when encountering an arterially hyperenhancing mass, where the differential diagnosis frequently includes hepatic adenoma, hepatocellular carcinoma, and fibrolamellar carcinoma (Table 24.1). Focal nodular hyperplasia tends to retain hepatobiliary agents to a similar degree or greater than the adjacent liver parenchyma, which is rarely found in the alternative lesions in the differential.25

Radionucleotide imaging of focal nodular hyperplasia with technetium sulfur colloid (Tc-SC) is rarely performed today, but was used in the past to help differentiate focal nodular hyperplasia from other lesions, such as hepatic adenoma or hepatocellular carcinoma. Tc-SC with particle sizes of 0.3 to 1.0 µm are taken up by Kupffer cells within the liver. Focal nodular hyperplasia typically contains Kupffer cells and will demonstrate Tc-SC activity greater than the background liver in 40%, similar in 30%, and less in 30%.26 By comparison the majority of hepatic adenomas and hepatocellular carcinoma appear as a photopenic defect within the liver because they have less Kupffer cells than the background liver.

Figure 24.6 Focal nodular hyperplasia. (A) US image with Doppler flow overlay showing a hypoechoic mass (arrowheads). A feeding artery (arrow) is seen coursing into the center of the mass and smaller arteries are seen radiating from the center to the periphery of the mass (dashed arrows). This is the classic “spoke wheel” vascularity of an FNH. (B) Axial fat-suppressed T2-weighted MRI image shows the same mass (arrow) as T2 hyperintense to the remaining liver parenchyma with a centrally markedly T2 hyperintense central scar (arrow head). (C, D) Axial fat-suppressed T1-weighted images enhanced with hepatobiliary agent in the late arterial (C) and 20-minute delayed phases (D). Late arterial phase image shows diffuse avid contrast enhancement (C, arrow) with a T1 hypointense central scar (C, arrowhead). 20-minute delayed phase shows retention of contrast agent (D, arrow) to a degree greater than the surrounding liver, which is typical of an FNH. The central scar does not retain hepatobiliary contrast agent (D, arrowhead).

Radionucleotide imaging with the hepatobiliary agent technetium iminodiacetic acid (Tc-IDA) analogs show uptake in normal hepatocytes and excretion into the biliary tree. When focal nodular hyeprlasia is imaged with these agents, rapid uptake and delayed clearance are seen in over 90% of patients.18 By contrast, adenomas do not take up the agent, whereas hepatocellular carcinomas often show only some delayed retention, although this is absent in poorly differentiated tumors.

Table 24.1 Comparison of common hyperenhancing liver masses


Imaging characteristics

Attenuation at CT

T1 Signal

T2 Signal





Marked Hyperintensity

Peripheral globular discontinuous enhancement with gradual complete central filling. Regions of contrast enhancement should appear similar to the aortic blood pool

Focal nodular hyperplasia

Isoattenuating with hypoattenuating central scar

Isointense to slightly hypointense

Isointense to slightly hyperintense

Diffuse late arterial hyperenhancement with fade to isoenhancement on delayed phases


Generally heterogeneously hypoattenuating with regions of hemorrhage appearing hyperattenuating and lipid-containing regions hypoattenuating

Generally heterogeneously hypointense with regions of hemorrhage appearing hyperintense and lipid-containing regions hyperintense (hypointense if fat saturation applied)

Hyperintense. Lipid rich lesions may appear hypointense on fat saturated sequences.

Heterogeneously hyperenhancing in the late arterial phase with fade to isoenhancement on delayed phases. Delayed-phase hypointensity (washout) may be present.

Regenerative nodule




Iso to slightly hypoenhancing

Dysplastic nodule

Isoattenuating or slightly hyperattenuating

Isointense to slightly hyperintense


Iso to hyperenhancing in the late arterial phase with fade to isointensity on delayed phases

Hepatocellular carcinoma

Variable. Generally hypoattenuating but often heterogeneously isoattenuating or hyperattenuating because of the presence of iron, hemorrhage and lipid.

Generally hypointense with regions of hemorrhage appearing hyperintense and lipid-containing regions hyperintense (hypointense if fat saturation applied)

Moderately T2 hyperintense but variable depending on necrosis, hemorrhage, iron, lipid content.

Hyperenhancing in the late arterial phase with hypointensity on portal venous or delayed phases (washout). Enhancement is often heterogeneous.

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Nov 24, 2019 | Posted by in GASTROENTEROLOGY | Comments Off on Imaging of Liver Tumors

Full access? Get Clinical Tree

Get Clinical Tree app for offline access