Multidetector computed tomography (MDCT) is a widely available imaging technique capable of providing, in a short examination time, images characterized by excellent spatial and contrast resolution. The MDCT protocol should include both unenhanced and enhanced scans, as the former, obtained at baseline, can be useful in the detection of intralesional calcifications (which may occur in F- and NF-PanNENs) (Fig. 9.4), and to accurately plan the dynamic contrast-enhanced study.

The intravenous administration of iodinated contrast agent is needed to optimally visualize the pancreatic parenchyma, increasing contrast resolution [28].

PanNENs are frequently hypervascular focal lesions, appearing as high-attenuating lesions during early contrast-enhancement phases [29, 30] (Figs. 9.5, 9.6). Accordingly, the protocol should include at least two contrast-enhanced phases [30–32], acquired, respectively, with a 40- to 45-s delay (arterial pancreatic phase) and a 70- to 80-s delay (portal venous phase) after the administration of contrast material (calculated by using the bolus-tracking technique).

However, about 30% of these tumors will have an atypical vascular pattern, resulting in iso- or even hypoattenuating lesions with respect to adjacent pancreatic parenchyma (Fig. 9.7) [13].

In our experience, in most cases the best enhancement is obtained during the arterial pancreatic phase [13]; nonetheless, additional contrast-enhanced scans should be taken in selected cases. An early arterial phase, acquired with a delay of 20–25 s (vascular arterial phase), may be useful in the detection of small tumors characterized by subtle brief enhancement. In addition, it may allow detailed arterial vascular mapping, which is useful for staging locally advanced tumors and for treatment planning (Fig. 9.8).

A late venous phase, acquired with a delay of about 120 s, may be useful for depicting a delayed hypervascular enhancing pattern. A multi-phase imaging protocol therefore offers the advantage of increasing the possibility of demonstrating the typical hypervascular pattern of PanNENs, to allow locoregional staging and the detection of liver metastases [13].

Curvilinear reconstructions should be used to highlight the relationship between primary tumors and the pancreatic and biliary ductal systems (Fig. 9.9).

The patient can be administered a glass of water immediately before the examination, to assure optimal filling of the stomach and duodenum with a low-contrast medium and for better definition of the gastric and duodenal wall. We usually avoid administrating oral iodinated contrast material, to avoid the misinterpretation or masking of hypervascular lesions ectopically located within the duodenal or small-bowel wall.

Baseline CT usually depicts PanNENs as isodense masses with respect to normal parenchyma (Figs. 9.5a, 6a, 7a). Thus, the tumor might be missed on an unenhanced scan, unless it has a large diameter and/or is associated with a distorted morphology of the gland [13].

Tumor inhomogeneity may be caused by globular or lamellar calcifications, seen both in F-panNENs (mostly insulinomas) and in NF-PanNENs, in these cases often associated with large areas of necrosis (Fig. 9.4a, b, d) [33].

During dynamic contrast-enhanced study, both functioning and non-functioning PanNENs usually appear as well-defined round or oval-shaped hypervascular masses [33]. At MDCT, the most frequent peak-enhancement phase is the arterial pancreatic phase whereas rapid washout is frequently seen in the portal venous and late venous phases (Fig. 9.6).

Achieving an intense enhancement is also useful to better define the dimensions of the tumors and to evaluate the relationship with adjacent structures [34] (Fig. 9.10).

Enhancement is usually homogeneous in small lesions, measuring < 3 cm in diameter [13, 33]. Conversely, large tumors, measuring > 3 cm in diameter, are typically inhomogeneous because of the presence of intralesional necrotic areas or cystic degeneration [13, 35], appearing as hypodense areas compared with the viable hypervascularized neoplastic tissue (Fig. 9.9).

MDCT, due to its high spatial, contrast, and temporal resolution, is advantageous in loco-regional staging, depicting the encasement of both the arterial (superior mesenteric artery or the celiac axis) and venous (superior mesenteric vein and portal vein) vessels [13, 33]. Three-dimensional reconstruction of the peri-pancreatic vessels may be of help in treatment planning (Fig 9.8).

Neoplastic thrombus within the peri-pancreatic veins has the same density as the mass from which it derives, thus displaying a slightly lower density than the vascular lumen.

Secondary phenomena, such as bile duct dilation or vascular encasement, may be well-demonstrated by means of MDCT, using axial native images and multi-planar dedicated reconstructions (Fig. 9.9b) [13]. Also, dilatation of the biliary tree may be depicted in case of tumors located in the pancreatic head.

Depending on its location, the primary tumor may dislocate or compress adjacent structures such as the stomach and duodenum, spleen and left kidney, and the adrenal gland (Fig. 9.11). Frank invasion into adjacent viscera is rare and is usually associated with the presence of an endocrine carcinoma.

Even if less accurate than MRI in identifying liver metastases, MDCT represents a reliable tool for the identification of metastatic involvement of the liver in case of PanNENs [13, 33].

There is no difference between the metastases of F-PanNENs and NFPanNENs. These lesions usually share imaging features of primary tumors, i.e., slightly hypodense compared with the normal parenchyma on unenhanced CT scan, and hyperdense hypervascular lesions (sometimes with a target-like pattern) on arterial enhanced scan [30, 36]. Liver metastases typically show washout during portal venous and late venous phases, resulting in lesions hypodense to normal liver parenchyma (Fig. 9.3c, d). Calcifications may be present as well.

Hypovascular liver metastases may be associated with hypovascular primary tumors (Fig. 9.7).

The typical features of NF-PanNENs, such as a well-demarcated hypervascular mass with a compressive pattern of growth, are present in about 70% of patients [13, 15, 28, 33]. In the other 30%, the pattern is non-specific and a reliable differential diagnosis with ductal adenocarcinoma is not possible [37], since the tumor is mainly hypodense compared with the pancreatic parenchyma (Fig. 9.7).

When the mass is large and well-circumscribed, a ductal adenocarcinoma can be excluded, but the problem of differential diagnosis from other rare, solid tumors remains, including solid variants of micro-cystic cystadenoma and pancreatic metastases.

PanNENs may appear as iso-attenuating to normal pancreas in pancreatic phase CT images, and sometimes are better delineated in the portal venous phase [13, 38]. Late enhancement of the tumor may be explained by extensive necrosis, resulting in a slower washout from the mass, due to the reduced vascularization [39].

In patients with PanNENs, the information obtained with MRI is similar to that obtained with CT, with the additional advantage that the patient is spared radiation exposure. However, due to its relatively limited availability and the longer examination time, MRI is not as widely used as CT for imaging pancreatic tumors.

State-of-the-art MRI of pancreatic neoplasms is optimally performed with 1.5 Tesla gradient systems using phased-array coils to improve the signal-tonoise ratio, optimized with thin slices and a small field of view [13, 40]. Breath-hold acquisitions are obtained with fast spin echo (FSE) or gradient echo (GRE) sequences and echo planar imaging. A moderately T2-weighted FSE and single-shot FSE (SSFSE) should be obtained, followed by T1-weighted in-phase GRE and T1-weighted opposed-phase GRE.

T1-weighted images with fat suppression have proven to be useful for imaging the pancreatic gland, allowing high contrast resolution between the normal bright parenchyma and the surrounding hypointense retroperitoneal fat [41]. Coronal and axial magnetic resonance cholangiopancreatography (MRCP) with SSFSE accurately depicts the pancreatic ducts.

For the evaluation of PanNENs, fat-suppressed three-dimensional spoiled GRE sequences after the administration of gadolinium-DTPA are acquired in arterial phase (30–40 s), portal phase (70–80 s), and equilibrium phase (180 s) [13, 40, 41]. The acquired images should cover the upper abdomen, including the entire liver, thus improving the detection and characterization of locoregional lymph-nodes and hepatic lesions [13].

Similar to contrast-enhanced helical CT, additional gadolinium enhanced scans can be obtained in the early arterial (scan delay 20 s) or late venous (120 s) phases, which increases the possibility of imaging the typical hypervascular enhancement pattern of these tumors [13, 31, 42].

On T1-weighted images, the normal pancreas exhibits medium to high signal intensity, similar to or slightly less than that of liver, but lower than that of retroperitoneal fat.

Fat suppression should be used, especially for T1 sequences, to increase pancreatic conspicuity, since with this technique the pancreas assumes a bright signal intensity that facilitates the detection of focal lesions [41, 42]. In patients with fatty involution of the pancreas, the signal intensity of this organ increases on T1-weighted sequences according to the amount of fat present within the parenchyma. Consequently, the pancreatic bed appears as an area of very low signal intensity on fat-sat sequences, which therefore are of little use in the visualization of small tumors (Fig. 9.12).

T2-weighted images with fat suppression may be useful for the evaluation of peri-pancreatic structures and inflammatory changes, but they are not strictly needed for imaging panNENs. On TSE T2-weighted images, the pancreas demonstrates intermediate signal intensity, similar to that of the liver. The signal may be intermediate to low on HASTE sequences.

The surface of the normal pancreatic parenchyma may be either smooth or lobulated.

Pancreatic ducts appear as low-signal intensity tubular structures on T1- weighted sequences and as high-signal intensity structures on heavily T2- weighted scans. At MRCP, heavy T2-weighting and fat suppression provide a cholangiogram useful for the evaluation of pancreatic and biliary duct involvement.

The typical MRI features of panNENs include a pancreatic mass of low signal intensity on T1-weighted images and of intermediate to high intensity on T2-weighted images (Fig. 9.12). As previously stated, the better intrinsic contrast resolution of this imaging technique may be advantageous for the identification of very small primary tumors, which frequently appear hypointense relative to the normal parenchyma on T1-weighted sequences. Lesion conspicuity is usually enhanced by fat-suppression.

Small lesions, which account for the majority of F-panNENs and some incidentally detected NF-PanNENs, are often quite homogeneous. Conversely, larger tumors may appear markedly inhomogeneous due to intralesional necrosis or hemorrhage, which may be seen as hyperintensity on T1-weighted images [43] and inhomogeneous hyperintensity on T2-weighted images (Fig. 9.13).

Cystic tumors have been described [35] and are often associated with widespread intralesional necrosis. Cystic lesions are usually unilocular, with contents that are hypointense on T1-weighted and hyperintense on T2-weighted images [44]. The cystic wall may show variable thickness [44]. The appearance of these tumors may be similar or even identical to that of other cystic tumors of the pancreas. Sometimes, intense enhancement of a peripheral ring-shaped viable tumor will suggest the diagnosis of cystic NF-panNENs (Fig. 9.14); however, in the majority of cases, a definitive diagnosis can only be obtained by histological examination of the resected specimen.

Among their atypical features, some islet tumors may have a low signal intensity on T2-weighted images due to the presence of abundant fibrous tissue [45]; in such cases they may be indistinguishable from ductal adenocarcinoma.

As seen for CT, during dynamic contrast-enhanced study, the majority of PanNENs show a typical hypervascularity [13, 31, 33, 46], resulting in hyperintense lesions compared to normal pancreatic and liver parenchyma (Figs. 9.11c, d, 9.12, 9.13).

In general, small tumors are depicted as homogeneously enhancing lesions (Fig. 9.12), whereas large tumors may appear markedly inhomogeneous during dynamic studies, since central necrotic areas remain hypointense on T1- weighted images even after contrast medium administration (Fig. 9.13).