Liver : Normal Anatomy, Imaging Techniques, and Diffuse Diseases

Surface Anatomy.

The liver's external structure features convex diaphragmatic and concave visceral surfaces. The diaphragmatic surface is perfectly fitted and affixed by intervening connective tissue to a triangular section of the undersurface of the diaphragm termed the bare area. Other than the bare area, only the fossa of the gallbladder, the fossa of the inferior vena cava (IVC), and the suprarenal impression are not covered by peritoneum. The bare area's upper and lower margins are demarcated by peritoneal reflections forming the anterior and posterior layers of the coronary ligament, respectively. The posterior layer of the coronary ligament continues with the right layer of the lesser omentum as the hepatorenal ligament. The anterior and posterior layers of the hepatic coronary ligament converge on the right and left margins of the bare area, forming the right and left triangular ligaments. The right triangular ligament is a small peritoneal fold that extends toward the diaphragm. The left triangular ligament on the other hand is of considerable size and gives off connective tissue tracts extending to the diaphragm as well as portions of the falciform ligament. The main anterior portion of the left triangular ligament forms the left layer of the falciform ligament, whereas the right layer of the falciform ligament is a continuation of the anterior layer of the coronary ligament. The falciform ligament thereby forms a broad and thin peritoneal fold that dissects the liver into its right and left hepatic lobes and is in contact with the peritoneum posterior to the rectus abdominis musculature and the diaphragm.

Intraabdominal fixation of the liver to the undersurface of the diaphragm is therefore realized by intervening connective tissue of the bare area as well as by means of connective tissue tracts extending from the coronary and triangular ligaments; the rather lax falciform ligament attachment to the peritoneum only limits lateral displacement.

At the liver's visceral surface, predominantly fossae and fissures define the lobar anatomy. Two fossae extending sagittally are joined by a transverse fissure, forming a letter H –like structure. The left limb of the letter H is known as the left sagittal fossa, a deep groove extending from the anterior surface to the posterior contour of the organ, thereby marking the division of the liver into the right and left hepatic lobes. The anterior part of the left sagittal fossa houses the umbilical vein during fetal circulation and its adult remains, the ligamentum teres, thereafter. The ligamentum teres itself extends toward the base of the falciform ligament and runs together with the paraumbilical veins between its right and left layers from the umbilicus to the left portal vein. The posterior part of the left sagittal fossa contains the ductus venosus, which during fetal circulation forms a bypass between the left portal vein and the left hepatic vein, functioning as a left-sided portosystemic shunt, and during adulthood holds its remains, the ligamentum venosum. The anterior and posterior parts of the left sagittal fossa can be separated by the pons hepatis.

The right limb of the letter H consists anteriorly of the fossa for the gallbladder and posteriorly of the fossa for the IVC. The latter usually appears as a short and deep impression; occasionally liver parenchyma can surround the IVC and thereby form a complete canal. Three veins—the left, middle, and right hepatic veins—perforate the floor of the fossa and drain into the IVC. These two fossae are free of peritoneal lining and are separated by the caudate process of the liver.

The bar connecting the two limbs of the letter H is termed the transverse fissure. Anterior to the transverse fissure lies the quadrate lobe; posterior, the caudate lobe of the liver. The transverse fissure represents the porta hepatis, a 5-cm-deep groove containing the hepatoduodenal ligament, which transmits anteriorly the hepatic duct on the right and the hepatic artery on the left; posteriorly and partially in between the two resides the portal vein as well as hepatic nerves and lymphatics. The hepatic artery carries oxygenated blood to the liver, constituting about 20% to 25% of the total blood volume to the organ. The portal vein on the other hand carries nutrient-rich blood from the digestive system, comprising 75% to 80% of total hepatic blood volume. The hepatic artery and portal vein ramify in parallel to one another. In healthy persons, the caliber of the portal venous branches is larger than those of corresponding arterial branches. Venous return is realized through the three main hepatic veins that drain directly into the IVC.

The visceral undersurface of the liver is uneven owing to its proximity to the antral esophagus, stomach, right colonic flexure, upper pole of the right kidney, right adrenal gland, and descending portion of the duodenum. Each of the organs forms characteristic impressions. Additionally a rounded eminence, the tuber omentale, abuts the gastric concavity of the lesser curvature in front of the lesser omentum. The diaphragmatic surface can be riddled with furrows from prominent muscular bundles forming the diaphragmatic crura, seen particularly in patients with chronic lung diseases.

Functional Segmental Anatomy.

Advances in hepatic surgery have made the lobar segmentation of the liver into right, left, quadrate, and caudate lobes obsolete. This division has yielded to a more functionally oriented classification primarily based on the surgical definition of feasible intrahepatic boundaries of resection, which in turn is based on segmental hepatic configuration. Centrally located in each of the hepatic segments is a segmental branch of the portal vein and hepatic artery as well as a segmental bile duct. The solitary distal hepatic veins lie between the individual segments. Several systems have been suggested for classifying the segments of the liver, many of them quite similar; the most common is the nomenclature proposed by Bismuth-Couinaud, which will now be considered in detail.

In the Bismuth-Couinaud system, segment I corresponds to the caudate lobe. This segment is in the unique position of receiving branches from both the main portal vein and its right and left branches, termed the portal trinity. Furthermore, it does not drain into the hepatic veins but directly into the IVC. All remaining liver segments (II to VIII) are defined by their positions relative to branches of the portal and hepatic veins. The portal venous plane is defined as the transverse plane dissecting the liver at the level of the portal venous bifurcation into superior and inferior hepatic segments.

The portal vein divides early into left and right portal venous branches, with the left portal pedicle supplying segments II through IV, thereby forming the segmental left hepatic lobe, and the right portal pedicle (segments V through VIII) thus constituting the segmental right hepatic lobe.

The left portal vein, after giving off a caudate branch, divides into its terminal branches, the left lateral and left medial portal venous branches. The left lateral portal venous branch supplies superior segment II, located lateral to the left hepatic vein and above the portal venous plane. The left medial portal venous branch supplies inferior segment III located laterally to the left hepatic vein and beneath the portal venous plane as well as segment IV. Segment IV is delineated medially by the middle hepatic vein and laterally by the left hepatic vein and can be further subdivided into a superior segment IVa and an inferior segment IVb in regard to the portal venous plane.

The right portal vein divides into its right anterior and right posterior terminal branches. The right anterior portal venous branch supplies the anterior-inferior segment V located between the middle and right hepatic veins and below the portal venous plane, whereas the anterior-superior segment VIII can be found contiguous to segment V but directly above the portal venous plane. The right posterior portal venous branch perfuses the posterior-inferior segment VI located posterolaterally to the right hepatic vein but beneath the portal venous plane, while the posterior-superior segment VII can be found adjacent to segment VI directly above the portal venous plane ( Fig. 43-1 ).

Variants in Morphology.

Common normal variants in liver morphology include the horizontal elongation of the lateral segment (Bismuth-Couinaud segment II) of the left hepatic lobe, which can extend into the left upper abdominal quadrant and eventually abut or even wrap around the splenic contour. This anatomic variant is more common in women. However, because of the different computed tomography (CT), magnetic resonance imaging (MRI), and perfusion characteristics of hepatic and splenic parenchyma, this variant usually does not impose problems during radiologic interpretation.

Another normal variant of liver size and shape is vertical elon­gation of the right lobe, termed a Riedel lobe ( Fig. 43-2 ). The presence of a Riedel lobe usually results from a prominent inferiorly positioned narrow right lobe of the liver that significantly extends the expected confines of the liver. A Riedel lobe is more frequently found in women. It has to be differentiated from extracapsular extension caused by a liver tumor, in particular hepatic adenoma or metastasis.

A prominent caudate lobe can show a pronounced extension, which is then termed a hepatic papillary process ( Fig. 43-3 ). This normal variant can simulate a mass in the portocaval region. Differentiation between an extrahepatic portocaval mass and a papillary process can be facilitated by recognizing the continuity of this process with the caudate lobe on contiguous axial sections, as well as multiplanar reformations (MPR), preferably in the coronal plane.

Following a partial hepatectomy, rapid hypertrophy of the remaining liver segments can occur. This hypertrophy results in distortion of the expected anatomy as the major vessels become displaced by the enlarging segments. Furthermore, segmental hepatic resections have to be differentiated from congenital hypoplasia of the left lobe. Assessment of the clinical history of the patient and direct patient communication remain of paramount importance when interpreting complex anatomy.

Variants in Vascularity.

Common normal variants in hepatic vascularity have to be identified prior to implementation of the Bismuth-Couinaud system to describe the location of hepatic focal lesions. Prior knowledge of replaced hepatic arteries is of importance in clinical scenarios of planned partial hepatectomies and extensive pancreatic surgery. Interindividual frequency and degree of variability of the hepatic arterial and venous as well as portal venous vasculature, particularly in the right liver lobe, is significant. Variants in arterial hepatic supply exist as accessory or replaced right hepatic arteries, commonly originating from the superior mesenteric artery, then located posterior in regard to the portal vein. In up to 20% of all patients, a duplication of the right hepatic vein may be found. High-resolution imaging of the hepatic veins and the portal venous system has shown that only the anterosuperior and posteroinferior sectors are visualized without significant variation in vascularity. It was shown that portal venous branches often cross the plane of the right hepatic vein, and no portal venous plane can be identified separating the anterosuperior and posteroinferior sectors; hence the right hepatic vein cannot serve as a reliable landmark for dividing the anteromedial and posterolateral segments of the right hepatic lobe.

The left liver lobe presents fewer variations in vascular anatomy. Analogously, variants in arterial hepatic supply exist as accessory or replaced left hepatic arteries, commonly originating from the left gastric artery and then always coursing within the fissure of the venous ligament. Duplication of the middle hepatic vein is seen in about 5% of patients, of the left in about 15%. There may also be some variation in arterial structure, particularly in segment IV. In about 8% of all patients the dorsal portion of segment IV is supplied by the right hepatic artery, hence from the functionally contralateral side.

In about 30% of patients, accessory portal segments analogous to the caudate lobe (Bismuth-Couinaud segment I) may be observed, each of which is supplied directly from the main portal venous trunk or from its right branch.

Recognizing the significant interindividual variation in hepatic vascularity, advanced and less radical hepatic surgery aims to reduce the reliance on a rigid schematic division of the liver into segments and considers instead the patient's individual vascular situation in surgeries involving segments or even subsegments of the liver. A liver segment is defined solely by its vascular supply. Based on high-resolution multidetector (MD)CT and MR data sets, it is possible to construct three-dimensional (3D) hepatic maps of vascular anatomy to assist surgeons in exact and effective preoperative planning of the procedure.

MDCT Imaging Strategies.

The multitude of clinical scenarios that require or include hepatic MDCT imaging cannot be appropriately addressed with one comprehensive imaging protocol. Instead a manageable list of MDCT imaging protocols to answer specific clinical situations have been defined throughout medical institutions. Furthermore, each of these imaging protocols is tailored to the individual patient, taking individual descriptors and parameters such as body habitus, body weight, heart rate, and limitations in the patient's range of motion and positioning into account.

Valid indications for non–contrast-enhanced (NCE) hepatic MDCT examinations include (but are not limited to) the search for calcifications in settings such as hemangioendothelioma, calcified metastases of mucinous adenocarcinoma, and postinflammatory calcifications in cases of alveolar echinococcosis. Furthermore, unenhanced MDCT can help detect intrahepatic CT-opaque bile duct concretions. Subcapsular or parenchymal hemorrhage can also be visualized. Finally, postprocedural visualization of the lipiodol distribution pattern following intraarterial embolization of liver tumors is another valid indication for NCE hepatic MDCT imaging.

Whenever contrast-enhanced MDCT imaging of the liver is part of a more comprehensive imaging protocol—for example, including the entire chest, abdomen, and pelvis in the imaging field of view (FOV) without suspected specific hepatic pathology—usually only one hepatic contrast phase is acquired. In this contrast monophase imaging scenario, the portal venous imaging phase will be preferred. Dedicated multiphase hepatic imaging can be performed subsequently to further evaluate suspected hepatic findings.

In clinical scenarios with a known or suspected hypervascular primary neoplasm outside the liver for which there are suspected hypervascular hepatic metastases, a dedicated hepatic dual-phase imaging protocol should be performed. This protocol includes MDCT imaging during both the hepatic late arterial and the portal venous phase but does not include an NCE phase. Confirmed or suspected diagnoses of breast carcinoma, renal cell carcinoma, melanoma, and neuroendocrine tumors (e.g., islet cell tumors, carcinoid and thyroid carcinomas) would call for the dual-phase hepatic imaging protocol.

Known or suspected cirrhosis or hepatocellular carcinoma as well as suspected benign primary liver lesions (e.g., focal nodular hyperplasia or hepatic adenoma) would be imaged by means of MDCT employing a triple-phase imaging protocol. This protocol includes a noncontrast imaging phase in addition to hepatic arterial and portal venous imaging phases; it also calls for a larger volume of contrast agent and higher infusion rates compared to the dual-phase protocol. Oftentimes a delayed phase acquired 10 to 15 minutes after initiation of contrast is performed as well.

In patients with known or suspected cholangiocarcinoma another triple-phase imaging protocol (unenhanced, portal venous, and delayed) addresses a pathophysiologic phenomenon of delayed contrast agent washout and subsequent relative hyperattenuation of cholangiocarcinomas in comparison to surrounding hepatic parenchyma.

For preoperative planning in anticipation of a partial hepatic resection, high-resolution imaging of the celiac axis, proximal superior mesenteric artery, and porta hepatis is mandatory. A hepatic resection protocol combines high-resolution imaging of the arterial vasculature with a limited imaging FOV followed by a standard portal venous imaging phase.

MDCT Imaging Protocols.

The family of available CT imagers covers a huge bandwidth of detector configurations from single- to 128-detector arrays, which allow helical scanning at various temporal and spatial resolutions. Fundamental differences in detector configurations of equally sized and equally spaced detector arrays from asymmetric detector configurations of detector sizes and arrangements result in MDCT imaging protocols with unique parameters for almost each available MDCT imaging system. The advent of dual-source and/or dual-detector MDCT imagers that allow for simultaneous dual-energy examinations will increase possible combinations of imaging parameters even further.

This overview of MDCT imaging protocols attempts to cover a representative selection of widely used equally sized and equally spaced 16-detector and 128-detector MDCT images. Adaptations according to individual scanner configurations, however, can be performed using these imaging protocols as groundwork ( Tables 43-1 to 43-6 ) (see Fig. 43-1 ).

Dual-energy CT applications for liver imaging.

Dual-energy MDCT allows for spectral analyses and thus tissue characterization based on using two x-ray beams of distinct and known energy distributions tuned to the characteristic absorption profiles of different tissues and injected contrast materials. This has the potential for enhancing inherent tissue differences and enabling selective contrast material detection.

The identification, isolation, and quantification of iodine-containing voxels throughout the liver enable the creation of iodine maps from contrast-enhanced dual-energy MDCT studies (DECT). Alternatively, virtual nonenhanced image series might be obtained by subtracting the iodine contribution from the contrast-enhanced image data sets. The latter approach has the potential for substantially reducing the radiation dose to patients for multiphase MDCT protocols of the liver ( Fig. 43-4 ).

A more recent promising application for DECT imaging of the liver is based on material density analysis. Material density images are obtained using a two-material decomposition algorithm, which works by first establishing a basis pair and identifying associated attenuation characteristics of each material in the basis pair. Then each voxel is decomposed proportionally into the two-material basis pairs according to the attenuation characteristics of that specific voxel as measured at each energy level (typically 80 and 140 kVp). Material density images represent the relative distribution of densities in a given material, which produce the observed projections at the high and low energies. Therefore the values obtained on material density images do not represent absolute density measurements, but rather they reflect the true proportion or relative quantity of material under investigation. As an example, for a material basis pair composed of fat and iodine, the effective fat density image is represented by the fat-iodine data set, whereas the effective iodine density image is represented by the iodine-fat data set. A region-of-interest measurement obtained on a fat-iodine image data set represents the relative effective density of fat without the iodine contribution. Similarly, an ROI measurement drawn over an iodine-fat image depicts the relative effective density of iodine without fat within a given target area.

MRI Strategies.

MR sequences for hepatic imaging continue to evolve at a fast rate, whereas the clinical applications of each of these sequences have changed little over time. Unchanged remain the three basic considerations if MRI has been chosen for hepatic imaging: to improve parenchymal contrast, to suppress respiratory motion artifact, and to ensure complete anatomic coverage. Guaranteeing these three basic considerations in order to assess various clinical scenarios of differing hepatic diseases using the most advanced designs of T1-weighted and T2-weighted imaging sequences remains the greatest challenge in hepatic MRI.

A major recent development—particularly in abdominal MRI—was implementation of parallel imaging techniques using spatial information from single coils of a phased-array set to perform a portion of the spatial encoding normally accomplished by time-consuming gradients and radiofrequency (RF) pulses, thereby significantly reducing acquisition times. Different approaches such as the simultaneous acquisition of spatial harmonics (SMASH) technique or the sensitivity encoding (SENSE) procedure have been proposed, realizing spatial encoding with multiple spatial-distinct receiver coils for a sparse sampling of k-space. Subsequently, linear combinations of component coil signals are used to emulate the effects of phase encoding gradients.

However, basic parallel imaging techniques require detailed and time-consuming coil mapping procedures for calibration purposes to prevent image distortions. Faster and more sophisticated calibration techniques acquire a block of extra lines in the center of k-space and subsequently perform a sliding block reconstruction. Unlike the internal autocalibration functions of SMASH techniques (AutoSMASH), data from multiple lines from all coils are used to fit an autocalibration signal line in one single coil. This fit gives the weights that can be used to generate the missing lines from that coil. Once all of the lines are reconstructed for a particular coil, a Fourier transformation can be used to generate the uncombined image for that coil. This process is repeated for each coil of the array, and finally the full set of images can be combined using a sum-of-squares reconstruction. This approach is also known as the generalized autocalibrating partially parallel acquisition (GRAPPA) technique.

Acceleration of individual imaging sequences and use of respiratory triggering techniques (e.g., respiratory bellows or implemented diaphragmatic navigators) ensure that T1-weighted and T2-weighted pulse sequences can be acquired in excellent quality. Whereas T1-weighted imaging is currently performed during a single breath hold—particularly important for dynamic contrast-enhanced (DCE) imaging—T2-weighted pulse sequences are obtained in a seamless multisegmental end-expiratory fashion using respiratory triggering.

Developments and improvements in magnetic field strengths, gradient technology, and receiver coil design, as well as implementation of parallel imaging techniques and respiratory triggering procedures, allow acquisition of a comprehensive hepatic MR exam in less than 30 minutes. For hepatic MRI the following sequence types are used: single-shot fast-spin echo (SSFSE), balanced steady-state free precession (SSFP), dual gradient echo in-phase (IP) and opposed-phase (OP) imaging, dynamic multiphase 3D spoiled gradient echo, diffusion-weighted imaging (DWI), and MRI elastography.