Chapter 17 Magnetic resonance imaging of the liver, biliary tract, and pancreas

Magnetic resonance imaging (MRI) is a cross-sectional scanning technique that allows images to be taken in any plane (Fig. 17.1). MRI uses magnetic fields and radiofrequency pulses to generate images with outstanding tissue contrast and excellent spatial resolution. MR techniques are not new and were first described in the 1940s by Bloch and colleagues (1946) and Purcell and colleagues (1946) as a method for in vitro chemical analysis. Several decades later, Damadian (1972) and Lauterbur (1973) applied some of these basic principles to design an MRI technique capable of in vivo imaging. Today, MRI is used extensively as an imaging tool throughout the body to visualize and distinguish normal and pathologic tissue.

FIGURE 17.1 Multiplanar T2-weighted images through the liver in a patient with multiple hepatic metastases. A, Axial image. B, Sagittal image. C, Coronal image. A variety of techniques can be used to obtain these images. Note that fluid-containing structures, including small bowel (curved arrow, A), gallbladder, biliary tree, and pancreatic duct (arrow, C) are bright.

Principles of Magnetic Resonance Imaging

MRI is based on the principles of nuclear magnetic resonance. Certain nuclei have a magnetic movement or spin. These nuclei react to placement in a strong magnetic field by aligning themselves in the direction of the field. A proton aligns itself in one of two directions, either up or down. In addition, when placed in a magnetic field, the nuclei rotate or spin parallel to the magnetic field. The precise frequency of the nuclear spin is termed the Larmor frequency, which depends on the specific type of nuclei imaged within the magnetic field and the strength of the magnetic field. The most commonly imaged nucleus in clinical practice is hydrogen (¹H) because of its great abundance in the human body. Other nuclei that may be imaged by MRI include phosphorus (³¹P), sodium (²³Na), and carbon (¹³C).

When nuclei are within the stable magnetic field, they are considered to be in equilibrium: an almost an equal number of nuclei are in an “up” or “down” alignment, and an equal number of transitions exist between the two states. Adding a momentary radiofrequency pulse at the precise frequency of the nuclear spin (the Larmor frequency) produces resonance, hence the term magnetic resonance imaging. This radiofrequency pulse causes a change in the energy and a change in the transition state. The radiofrequency pulse is then turned off, and the nuclei in the body return to the equilibrium state, emitting a radiofrequency signal. The strength of this emitted signal and the amount of time for the nucleus to return to the equilibrium state determine the signal intensity of a tissue. The precise tissue signal intensity depends on several factors, including longitudinal relaxation (T1), transverse relaxation (T2), proton density (nuclear spin), and flow.

The T1 value, longitudinal or spin-lattice relaxation, is the time required for the rotating nuclei to return to the equilibrium state after the radiofrequency pulse. The initial energy absorbed by the nuclei is released and dissipated into the surrounding molecular environment or lattice. The precise T1 value is the time it takes the MRI signal to return to 63% of its maximum value. Most tissue T1 values vary between 200 and 800 ms. Tissue with a short T1 value appears bright (hyperintense) on T1-weighted images by convention.

The T2 relaxation time, transverse or spin-spin relaxation, is a measure of the loss of signal in a plane perpendicular to the long axis of the magnetic field. This loss of signal is due to subtle inhomogeneities in the magnetic field because of the presence of spinning protons. The T2 value for a tissue is the time it takes for the signal to decrease to 63% of its equilibrium value. Most tissue T2 values vary between 50 and 200 ms, and the T2 value can never be longer than the T1 value. By convention, tissues with a long T2 value appear bright on T2-weighted images.

The motion of fluid, such as from blood, plays a major role in the appearance of every MRI image. In general, blood flow tends to produce spin dephasing, which results in a loss of signal in an MRI image. Certain pulse sequences are sensitive to flow, and flow produces a higher signal intensity. MRI angiography techniques use this principle to produce images in which flowing blood is bright and background tissue is relatively dark. This technique takes advantage of the fact that blood flow produces moving protons that have a different signal from the static protons within the slice being imaged.

Improvements in MRI equipment and software used to control this machinery have resulted in a decrease in the time needed to create an image. State-of-the-art scanning equipment may create an image in less than 1 second. Also, specialized surface coils, local antennas used to receive the magnetic resonance signal, are now used to obtain images with higher spatial resolution. Fast techniques can produce images with either T1 weighting or T2 weighting. Gradient-recalled echo techniques can be applied to produce high-quality T1-weighted images covering the entire liver in a total scan time of less than 20 seconds. This fast scanning permits breath-hold imaging and allows for the dynamic administration of certain contrast material with hepatic MRI. New techniques using a half-Fourier technique—acquisition of slightly more than 50% of the data and the computer reconstruction of the remaining data set—create subsecond T2-weighted images. Although rapid T2 weighting has been accomplished previously with techniques such as echo-planer imaging, the half-Fourier techniques have significantly fewer artifacts. These new subsecond T2 sequences have several vendor-specific trade names, such as Single Shot Fast Spin Echo (SSFSE) or Half-Fourier Acquisition Turbo Spin Echo (HASTE). Despite different nomenclature, these techniques produce similar, highly effective results.

Diffusion MRI is also a new MRI method that produces images of tissues weighted with the local structural properties of water diffusion. Diffusion-weighted imaging (DWI) of the liver may be used to characterize focal hepatic masses and may also be used, for instance, to assess liver fibrosis. In DWI, each image voxel, a three-dimensional pixel, has an image intensity that reflects a single best measurement of the rate of water diffusion at that location. This measurement is more sensitive to early changes after a stroke than more traditional MRI measurements such as T1 or T2 relaxation rates. DWI is most applicable when the tissue of interest is dominated by isotropic water movement, for example, gray matter in the cerebral cortex and major brain nuclei, in which the diffusion rate appears to be the same when measured along any axis.

Magnetic Resonance Imaging Safety

MRI is safe for most patients. However, physicians and patients should be aware of some important considerations, as several contraindications to MRI examination exist. Absolute contraindications include a cardiac pacing device, implanted neurostimulator, intracranial aneurysm clip of unknown composition, and a metallic fragment in the globe of the eye. Relative contraindications for MRI include recent intravascular stent or filter and pregnancy. Depending on the location and composition of a particular stent or filter, a variable time frame exists before an MRI examination is obtainable. Pregnancy is a relative contraindication, because there are no known long-term effects of MRI on the human fetus. It is prudent, however, to avoid scanning a pregnant patient before completion of organogenesis. These contraindications are not all-inclusive, and if questions arise about the safety of the patient, it is best to consult with the MRI specialist performing the examination.

Magnetic Resonance Imaging Cholangiography

MRI cholangiography and MRI cholangiopancreatography (MRCP) are imaging techniques used to evaluate the biliary system. Heavily T2-weighted images are used to provide an overview of the biliary system and pancreatic duct. Excellent diagnostic-quality images are obtainable, with high sensitivity and specificity for evaluation of biliary duct dilation, strictures, and intraductal abnormalities (Sandrasegaran et al, 2010; Palmucci et al, 2010a; Palmucci et al, 2010b; Hekimoglu et al, 2008). Cross-sectional images and projection images (Fig. 17.2) may be produced easily with current MRCP techniques, and the projection images are similar to direct contrast-enhanced cholangiograms obtained with either endoscopic retrograde cholangiopancreatography (ERCP) or percutaneous transhepatic cholangiography (PTC; see Chapter 18).

FIGURE 17.2 Cross-sectional images and projection image of the same patient as in Figure 17-1. A, Cross-sectional coronal image obtained at a slice thickness of 4 mm. A focal mass (arrow) in the pancreas is hyperintense and composed of multiple cysts. B, Axial image through the pancreas also shows the mass. C, Projection of the abdomen obtained at a slice thickness of 40 mm. The cystic mass within the pancreas (curved arrow) is easily identified. Fluid-containing structures and lesions that act like fluid are easily seen with this technique. Note a small hemangioma at the hepatic dome (straight arrow).

The basic principle of MRCP is to use T2-weighted images, in which stationary or slowly moving fluid, including bile, is high in signal intensity; all the surrounding tissues, including retroperitoneal fat and the solid visceral organs, are lower in signal. MR-specific techniques for obtaining cholangiographic images include two-dimensional and three-dimensional sequences, breath-hold or non–breath-hold techniques, and respiratory gated techniques. MRCP plays an important role in imaging benign disorders of the biliary and pancreatic system, and it is part of a comprehensive imaging evaluation of malignancies of the biliary system (Kim et al, 2002b; Lee et al, 2003; Vaishali et al, 2004). MRCP is noninvasive, eliminating the morbidity associated with ERCP or PTC (Zhong et al, 2004). An additional advantage of MRCP includes visualization of the extrabiliary anatomy, allowing for exclusion or inclusion of alternative diagnoses. Surgical clips may create an artifact known as susceptibility, which may obscure the region of interest by producing areas of signal void. This artifact may mimic a stone, so caution must be used in evaluating MRCP images in postoperative patients to avoid a false-positive diagnosis.

Magnetic Resonance Imaging Contrast Agents

MRI contrast agents work by altering the T1 and T2 relaxation times of various tissue types. In the hepatobiliary system, MRI contrast agents can potentially improve the detection of liver lesions and improve the characterization of focal liver abnormalities. MRI contrast agents for hepatobiliary imaging are divided into two basic categories: intracellular and extracellular agents. The most commonly used contrast agents are the extracellular agents, such as gadopentetate dimeglumine (gadolinium diethylenetriaminepentaacetic acid [Gd-DTPA]), which is distributed within the intravascular compartment initially and rapidly diffuses through the extravascular space, similar to the action of iodinated contrast agents in computed tomography (CT; see Chapter 16). Agents such as gadolinium require fast MRI pulse sequences to preferentially enhance either tumor or normal tissue, so that the difference between the two tissues is great. Metastatic liver lesions enhance irregularly, predominantly in the periphery of the lesion, and slowly; the central portion may or may not accumulate contrast material. Using extracellular contrast agents such as gadolinium and dynamic fast scanning, hepatic hemangiomas can be differentiated from other tumors by their characteristic enhancement patterns. Hepatic hemangiomas fill in slowly (5 to 20 minutes) from the periphery with a nodular appearance (see Chapter 79A, Chapter 79B ). An advantage of gadolinium over iodinated CT contrast agents is the lack of renal toxicity even at high doses (Prince et al, 1996). This fact has clinical implications in patients with decreased renal function or in those at risk for potential nephrotoxic effects of iodinated contrast agents.

Several new hepatobiliary-specific contrast agents have been developed to overcome some of the limitations associated with the general extracellular contrast agents. These hepatobiliary-specific contrast agents are taken up to varying degrees by functioning hepatocellular tissue and are excreted at least partially in the bile. They function because of their paramagnetic properties, which cause T1 shortening both of the liver and biliary tree. Hepatobiliary-specific agents include mangafodipir trisodium, gadopentetate dimeglumine, and gadoxetic acid, and both gadobentate dimeglumine (Gd-BOPTA) and gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DPTA) have been approved in the United States. These contrast agents are all administered intravenously, although the dose and duration of the administration are different. These two agents are sometimes referred to as combination agents because of their dual capacity for imaging both in the dynamic phase and in the delayed, hepatocyte-specific phase. These agents have the potential to provide extensive and comprehensive information about the lobes of the liver, bile ducts, and liver vasculature. (Giovagnoni & Paci, 1996; Iman & Bluemke, 2000; Stern et al, 2000).

Normal Hepatic Appearance on Magnetic Resonance Imaging

Hepatic MRI is a sensitive technique to evaluate diffuse and focal abnormalities. In normal hepatic parenchyma on T1-weighted images, the liver is brighter (hyperintense) than the spleen. Depending on the specific technique chosen, the vessels may appear dark (flow voids on spin-echo technique), or they may be bright (gradient-echo technique). On T2-weighted images, the spleen is relatively brighter than the liver (Fig. 17.3). Fluid is dark on T1-weighted images and bright on T2-weighted images as a result of long T1 and T2 values. Although many hepatic lesions are low in signal intensity on T1-weighted images, they may be of variable intensity on T2-weighted images, depending on their water content.

FIGURE 17.3 Normal hepatic magnetic resonance image. A, T1-weighted spin-echo image. The liver is brighter (hyperintense) relative to the signal of the spleen. There are normal flow voids that appear as dark areas in the visualized vessels. Straight arrow, intrahepatic portion of the inferior vena cava; curved arrow, aorta. B, T2-weighted fast spin-echo image. Because of reversal of the liver/spleen contrast, the normal spleen is brighter than the liver. Note the flow voids within the vessels. Straight arrow, intrahepatic portion of the inferior vena cava; curved arrow, aorta. C, Postcontrast T1-weighted gradient-echo image. Normal enhancement in the liver and spleen is evident, and the parenchyma of both is about equal in this phase of the injection. All the vessels (straight arrow, intrahepatic portion of the inferior vena cava; curved arrow, aorta) are bright as a result of the T1 shortening effect of gadolinium.

Diffuse Hepatic Disease

Fatty Infiltration of the Liver

Fat may accumulate within hepatocytes for many reasons, including alcohol abuse, diabetes, drugs, and obesity (see Chapter 65). Fatty change may be diffuse, patchy, or focal. The pattern of fatty infiltration is related to regional differences in perfusion. It is often difficult to distinguish fatty infiltration from focal hepatic lesions on CT, because both appear low in attenuation. Additionally, focal fatty sparing may appear similar to a vascular neoplasm on contrast-enhanced CT. It is easy, however, to distinguish these entities on MRI because of different signal characteristics. On T1-weighted images, areas of fatty infiltration appear bright because of the low T1 value of fat. The appearance on T2-weighted images depends on the specific type of T2-weighted sequence acquired. Conventional spin-echo T2-weighted sequences are relatively insensitive to the presence of fat (Wenker et al, 1984). Areas of focal fatty infiltration may appear hyperintense to normal hepatic parenchyma on fast T2-weighted images because of the high signal intensity of fat in these sequences. Using chemically selective fat suppression techniques, these areas of focal fatty infiltration appear dark.

Chemical shift imaging is another technique that allows differentiation of the signals from fat and water protons. These chemical shift techniques are the most sensitive techniques for distinguishing fatty infiltration (Fig. 17.4; Mitchell et al, 1991; Siegelman, 1997; Springer et al, 2010). Chemical shift imaging techniques rely on the different resonant frequencies present in fat and water protons. Using fast imaging techniques, the signal emanating from fat and water protons may be equal but opposite. Tissues that have relatively equal quantities of fat and water appear dark, because the signals from fat and water cancel each other out. Areas of fatty sparing remain relatively hyperintense to the fatty infiltrated regions. Focal fatty infiltration should not enhance after gadolinium administration. A lesion with characteristic signal intensity findings on T1, T2, and gradient-echo opposed-phase imaging is diagnostic of fatty infiltration (Kreft et al, 1992; Mitchell et al, 1991). Any one of these sequences alone is not characteristic, because other lesions, including hepatocellular carcinoma (HCC) and adenoma, may contain small quantities of fat within them (see Chapter 78, Chapter 79A, Chapter 79B, Chapter 80 ).