MDCT is the single most important imaging tool to detect and to stage pancreatic cancer (Prokesch et al. 2003; Shrikhande et al. 2012; Varadhachary et al. 2006; Zeman et al. 1997; Pawlik et al. 2008). A localised tumour mass is the leading criterion in pancreatic cancer. Modern MDCT protocols use a dual-phase technique. 30 s after the intravenous injection of contrast, the arterial phase is acquired, and about 1 min after injection, the portal venous phase is taken. It is necessary to perform a rapid injection (3–5 mL/s) to maximise enhancement of the pancreas and mesenteric vasculature. MDCT has improved spatial and temporal resolution dramatically. The arterial phase images are required for visualisation of the primary tumour as well as evaluation of the arterial vasculature and its relationship with the tumour to predict resectability (Varadhachary et al. 2006). Pancreatic neuroendocrine tumours which are typically hypervascular can also be distinguished best in the arterial phase from PDAC. Compared to the surrounding normal pancreatic parenchyma, most PDACs are hypodense. In 5–10 % of PDACs, there is isoattenuation in both arterial and venous phase images. The subsequent venous phase is required to assess the liver and mesenteric venous vasculature as well as locoregional lymphadenopathy (Horton and Fishman 2002).

Most often water is used as a neutral oral contrast agent for visualisation of the stomach, the duodenum and the jejunum. Typically, thin collimation technique with slice thickness of around 0.6 mm is used with the reconstruction algorithm for 3–5 mm slices for standard interpretation. A second reconstruction at 0.5–0.75 mm is performed to create isotropic multi-planar reformations and 3D reconstructions. Often, two sets of 3D images are created: maximum intensity protection (MIP) and volume-rendered (VR) images. MIP imaging is invaluable for angiographic images to assess the relation of the tumour to the surrounding vasculature. VR allows optimal definition of the soft tissues, muscle, bone, and vasculature to demonstrate complex relationships between different structures at pathology.

For tumour detection, the sensitivity of MDCT is reported between 88 and 98 %. The smaller the tumours, the more difficult they are to visualise. PDACs tend to infiltrate dorsally into the retroperitoneum. Vessel involvement is of highest importance for the staging of local PDAC to classify pancreatic tumours into resectable, borderline resectable and non-resectable. One of the problems of the currently available literature is that discrepancies exist between the definitions of resectability, and therefore, it is highly recommended to consistently use the definition based on expert consensus statement published by Callery (Callery et al. 2009), which defines borderline resectable tumours as:

Tumour infiltration into the duodenum is important for the radiation oncologist because the duodenum is one of the dose-limiting organs for radiotherapy of pancreatic cancers. The loss of the fat plain between the mass and the duodenum, thickening of the duodenal wall, the extension of the tumour around the duodenum or duodenal obstruction are all signs of duodenal infiltration. A comparison between the tumour dimensions as measured on CT and pathology showed that the sensitivity of CT regarding duodenal invasion is quite low (Arvold et al. 2011). Unfortunately, endoscopic ultrasound is also not very sensitive to detect duodenal infiltration and cannot help to resolve the dilemma between tumour coverage and sparing of the duodenum as an important organ at risk.

Nodal disease is an important prognostic factor in pancreatic cancer, and the seventh classification of the UICC reflects this in defining stage IIB irrespective of T-category. Unfortunately, the ability of CT to distinguish malignant nodes is limited. It is purely based on size threshold of short axis ≥1 cm (Zeman et al. 1997). Correspondingly, the sensitivity of CT for positive nodes may be as low as 22 % (Pawlik et al. 2008). CT is well suited to identify metastatic disease to the liver which is best detected during the portal venous phase.

Advantages of MRI over CT are mainly based on the superior tissue contrast (Vachiranubhap et al. 2009). Due to the rich desmoplastic reaction characterised by fibrous stroma, PDAC is good visualised on fat-suppressed T1-weighted images. Tumours tend to be T1 hypointense in comparison with the relatively bright pancreatic parenchyma which avidly enhances on arterial phase images. With time, these tumours tend to increase enhancement on venous and delayed images. Therefore, MRI can be helpful to detect small tumours which can be difficult to visualise on CT especially below a size of 2 cm (Park et al. 2009; Fusari et al. 2010). An important indication for the use of MRI is isoattenuating pancreatic cancer which occurs in about 5–10 % of all pancreatic tumours. In these cases, MRI is an important tool to define the gross tumour volume (GTV) for radiotherapy because of its superior soft tissue contrast (Kim et al. 2010). A comparison between tumour sizes on abdominal MRI with pathologic specimen in resected tumours in 92 patients in 3D VIBE MRI showed that MRI underestimated tumour size by a median difference of 4 mm (range, −34 to 22 mm) (Hall et al. 2013). This is useful information for the expansion of GTV to clinical target volume (CTV). Additionally, alternative sequences can be useful. One of these is magnetic resonance cholangiopancreatography (MRCP) which is helpful to show subtle ductal narrowing (Miller et al. 2006). Diffusion-weighted images (DWI) measuring the random motion of water molecules within tissue render apparent diffusion coefficient (ADC) values. The motion of water is limited within tumours, and therefore, low ADC values can help to identify small lesions (Wang et al. 2011).

Compared to CT, enlarged lymph nodes are more conspicuous on MRI, especially on T2-weighted fat-suppressed sequences. Also, lymph nodes in the mesenteric fat are well visualised on non-fat-suppressed T1-weighted images because in this sequence lymph nodes appear in lower signal compared to the high-signal fat. MRI is particularly good in evaluating hepatic metastasis especially in small lesions where MDCT has limitations. Metastases are slightly hypointense on T1-weighted images and mildly hyperintense on T2-weighted images. Most lesions have peripheral enhancement on post-contrast images and wedge-shaped peripheral enhancement in the arterial phase (Miller et al. 2006).

Positron-emission tomography (PET) imaging without a contrast-enhanced CT component is not of use for screening purposes. However, PET-CT with IV contrast may help differentiate benign and malignant pancreatic lesions. The positive predictive value of PET-CT is up to 91 %, but the negative predictive value of only 64 % is problematic. Therefore, there is a high propensity for false-negative findings. False-negative findings are often observed as a consequence of hypoglycaemia or diabetes. The resolution of PET is best for lesions that are larger than 1 cm in diameter below which there is a high likelihood of false-negative results (Grassetto and Rubello 2011). Another situation where fluorodeoxyglucose (FDG)-PET imaging can be very helpful for tumour delineation is when the tumour infiltrates into the duodenum. Specifically, if patients have had a gastric bypass preventing appropriate oral contrasting of the duodenum, it can be very difficult to appropriately visualise tumour extension, and PET-CT is able to better discriminate duodenum from tumour.

The sensitivity and specificity of PET-CT for locoregional nodal disease have been reported to be only 46 and 63 %, respectively (Pakzad et al. 2006; Kauhanen et al. 2009). The most important reason to explain this poor performance is the insufficient resolution for lesions smaller than 1 cm. However, PET-CT plays a more important role for the detection of distant metastatic disease when combined with contrast-enhanced CT (Table 7.2). One study reported a change in resectability status in 21.5 % of the patients and change of the pretreatment stage in 26.9 % of the patients (Bang et al. 2006).

Evaluation of response to therapy in patients with locally advanced pancreatic cancer is not satisfactory with MDCT. Regression of primary tumours is rarely observed due to the rich desmoplastic reaction in pancreatic tumour masses which cannot be discriminated from viable tumour. PET-CT is superior to MDCT in that respect because it is evaluating viable tumour cells and can discriminate a scar from a true tumour mass (Grassetto and Rubello 2011). However, one pitfall is an inflammatory reaction appearing as a false-positive finding, a typical example for this is cholangitis. A small study in 15 patients before and after chemoradiotherapy showed that tumour response was observed more frequently at PET imaging compared to CT imaging (Bang et al. 2006).