Sequence design

An adequate sequence design of diffusion-weighted sequence in the upper abdomen is critical, because it has intrinsically limited spatial resolution. Box 1 summarizes the list of scanning parameters to be optimized in a DWI sequence of the body. Most commonly, a single-shot spin-echo echo-planar imaging sequence is performed, which has the advantage of a very fast readout, making it insensitive to macroscopic patient motion. However, this family of sequences is prone to motion and magnetic susceptibility artifacts. The maintenance of echo time (TE) as short as possible, using parallel imaging, high bandwidth, and advanced suppression techniques, minimizes distortion artifacts, although diffusion requires intrinsically long TEs due to the time required to impart sufficient diffusion sensitivity. In order to reduce the effects of respiratory and cardiac motion, it is necessary to use gated acquisitions ( Fig. 1 ). Most commonly, breath-hold or free-breathing sequences are used ( Box 2 ). In addition, fat suppression must be used to avoid ghosting artifacts from the fat signal.

Techniques of respiratory synchronization in DWI of the liver. Notice different SNR and acquisition time in ( A ) breath-hold (acquisition time: 120 seconds; 20 seconds per breath-hold), ( B ) free-breathing (acquisition time: 80 seconds), and ( C ) respiratory trigger (acquisition time: 220 seconds) acquisitions.

Modeling of diffusion signal

In order to use DWI as an oncological biomarker, quantitative mapping is critical, especially in the setting of therapy monitoring. The most widely used quantitative property is the ADC from a monocompartmental model of diffusion signal decay. ADC measurement minimizes the so-called T2 shine-through, referring to the high signal from long T2 species seen in DWI because of superimposed T2 weighting and permits evaluation of isolated diffusion effects.

However, the ADC model does not distinguish between the different compartments where the water protons can move, such as intravascular, extravascular, extracellular, and intracellular spaces. If several b values are acquired less than and greater than 100 s/mm ² , it is possible to differentiate between the fast movement of intravascular water molecules with low b values (<100 s/mm ² ), and the slow signal decay of diffusion signal with b values greater than 100 s/mm ² , secondary to restricted water movement in the extracellular and intracellular compartments (see Fig. 1 in the article by Broncano in this issue). This model is known as Intra Voxel Incoherent Motion (IVIM) because it has been found useful in the characterization of focal liver and pancreatic lesions, with advantage over ADC measurements in some scenarios. The contribution of true diffusion and perfusion toward signal loss is separated in this model and reflected in the following parameters: f (perfusion fraction) that represents the flowing water molecules within the capillaries; D (tissue diffusivity), a more reliable marker of tissue diffusion than ADC in organs with tissues with significant perfusion fraction; and finally, D* (pseudodiffusion coefficient), related to the perfusion contribution to signal decay.

If ultrahigh b values greater than 1500 s/mm ² are acquired, the remaining diffusion signal is related to a layer of polarized water molecules near of the charges of the membranes. The measurement of this very slow diffusion pool requires the use of a non-Gaussian model, such as diffusional kurtosis imaging (DKI), which reflects the microstructural complexity of tissue (see Fig. 1 in the article by Broncano in this issue). Derived parameters from DKI can help in the characterization of focal liver lesions. Compared with ADC, these models provide supplementary information of the diffusion signal from other compartments different to extracellular one ( Fig. 2 ).

DWI of HCC analyzed with different models of signal decay. ( A – C ) Precontrast, arterial, and venous phases of DCE-MR imaging show a large focal liver lesion with heterogeneous wash-in during arterial phase and posterior washout. ( D ) DWI with b value of 2000 s/mm ² shows high SI of the lesion and ( E ) very low ADC value (0.65 × 10 ⁻³ mm ² /s). IVIM model confirms the restriction of water diffusion of the lesion in the ( F ) D map with a value of 0.6 × 10 ⁻³ mm ² /s. ( G ) Increased perfusion fraction (16%) and ( H ) pseudodiffusion value (D*: 28.8 × 10 ⁻³ mm ² /s) demonstrate increased flow within the nodule. ( I , J ) Derived parameters from DKI show a kurtosis value of 1.1 and mean diffusion kurtosis of 1.4 × 10 ⁻³ mm ² /s.

b-values selection

ADC calculation minimally requires the use of 2 b values, although the more b values obtained, the better the reliability of ADC maps. In the abdomen, typically the ideal higher b value ranges between 600 and 1000 s/mm ² in order to maintain a sufficient signal-to-noise ratio (SNR). DKI requires a maximum b value in the range of 1500-2000 s/mm ² , and IVIM analysis requires several b values less than and greater than 100 s/mm ² , although it is feasible even with only 3 to 4 b values. However, the optimum set of b values and diffusion model used for analysis for oncological applications in the body is still to be defined.

Liver

Lesion detection

DWI using a low b value around 50 s/mm ² , also known as black-blood diffusion, allows better detection of small focal liver lesions against a liver parenchyma without vessel signal ( Fig. 3 ). This type of approach has been shown to improve detection of metastases in comparison to fat-suppressed T2-weighted sequences and is of special interest for identifying metastases from colorectal cancer before surgery, with particular advantage in the detection of lesions smaller than 1 cm and those adjacent to vascular structures. Furthermore, in patients at risk for nephrogenic systemic fibrosis, DWI can be considered an alternative to contrast-enhanced imaging. DWI in combination with gadoxetic acid (Gd-EOB-DTPA) improves detection of liver metastases compared with any individual technique alone, even after neoadjuvant chemotherapy.

Breast cancer liver metastasis ( arrows ) detection. ( A ) Axial turbo spin-echo (TSE) T2-weighted sequence does not depict a lesion in segment IVa, which was hardly visualized in ( B ) postcontrast portal phase of DCE-MR imaging sequence ( arrow ). ( C–E ) Axial DWI with b values of 0, 50 and 900 s/mm ² : notice how the lesion is better visualized ( arrows ) against a background liver parenchyma without vessel signal with b value of 50 s/mm ² . High SI on high b-value image suggests its malignant origin.

Lesion characterization

The role of DWI in focal liver lesion characterization is controversial due to misregistration between images with different b values, caused by respiratory motion. Adequate reproducibility has been achieved for ADC, but lower for f and D∗. DWI has been shown to accurate differentiate between cystic and solid liver lesions. Benign lesions tend to show higher ADC values and lower SI with high b values than malignant ones ( Fig. 4 ). However, there are no definitive data to support the use of ADC in the distinction between benign and malignant focal liver lesions, due to substantial overlap, particularly focal nodular hyperplasia (FNH) and adenomas with hepatocellular carcinoma (HCC) and metastasis. In addition, radiologists must be alert to potential pitfalls such as low ADC values seen in pyogenic abscess or high ADC values of necrotic or mucinous metastasis.

Typical behavior on DWI of the most common focal liver lesions ( arrows ). Paired high b-value image and corresponding ADC map of ( A , B ) Simple cyst: ADC value: 2.1 × 10 ⁻³ mm ² /s. ( C , D ) Hemangioma: ADC value: 1.4 × 10 ⁻³ mm ² /s. ( E , F ) FNH ADC value: 1.5 × 10 ⁻³ mm ² /s. ( G , H ) Hepatocellular adenoma: ADC value: 0.9 × 10 ⁻³ mm ² /s. ( I , J ) HCC: ADC value: 0.65 × 10 ⁻³ mm ² /s. ( K , L ) Colorectal liver metastasis: ADC value: 0.8 × 10 ⁻³ mm ² /s. ( M , N ) CHC: ADC value: 0.9 × 10 ⁻³ mm ² /s. Notice the overlap in ADC values between benign and malignant solid lesions.

IVIM-derived D and ADC show inconclusive results in the distinction of benign and malignant lesions, although with contradictory data about which of both parameters is more accurate. D* and f are useful to differentiate hypervascular from hypovascular lesions.

Therapy monitoring

DWI has been used for early assessment of tumor response to treatment of both primary and secondary malignancies of the liver ( Figs. 5 and 6 ). The good reproducibility of ADC measurement means that any ADC changes after treatment can be related to treatment effects. In general, increases in ADC after 1 week of successful treatment can be detected in the liver.

Therapy monitoring of breast cancer liver metastasis ( arrows ) with DWI. ( A , B ) Pretreatment DWI with b value of 900 s/mm ² and corresponding ADC map show lesions with true restriction of water motion in segment IVa. ADC value: 0.48 × 10 ⁻³ mm ² /s. ( C , D ) 4 months after chemotherapy DWI and ADC map demonstrate partial response with decrease in size and in SI in high b-value image, with increase in ADC value up to 1.1 × 10 ⁻³ mm ² /s.

Therapy monitoring of colorectal cancer liver metastasis ( arrows ) with DWI. ( A , B ) Pretreatment DWI with b value of 900 s/mm ² and corresponding ADC map show a lesion with heterogeneous restriction of water motion in segment VIII. High pretreatment ADC value of 1.5 × 10 ⁻³ mm ² /s suggests poor response to treatment. ( C , D ) 8 months after chemotherapy DWI and ADC map demonstrate progression with increase in size and in SI on high b-value image and decrease in ADC value: 0.73 × 10 ⁻³ mm ² /s.

For example, DWI with ADC is useful in the early assessment of HCC response after transcatheter arterial chemoembolization (TACE), and differentiating between viable and necrotic portions of the treated HCCs. Furthermore, DWI can detect recurrent tumor after treatment and predict HCC response to TACE using the monoexponential or biexponential models of analysis.

DWI also can be used as a biomarker of treatment response of HCC to the antiangiogenic agent, sorafenib. First, ADC decreases probably related to hemorrhagic necrosis and, posteriorly, increases because of tumor necrosis. A delayed ADC decrease suggests tumor recurrence. IVIM-related D and f have been shown also to be valuable markers of treatment response to sorafenib for HCC.

For therapy monitoring of hepatic metastases, increases of ADC after the start of chemotherapy or radiotherapy have been related to responding metastases, although there is no defined threshold of increase in ADC to define response. These changes can be detected during the first week after the start of treatment in colorectal and breast liver metastases and occur before changes in size. In addition, DWI appears superior to positron emission tomography/computed tomography (PET/CT) for early response assessment of metastases of common solid tumors treated with Y90 radioembolization. Lower pretreatment ADC value has been related to good response to chemotherapy in colorectal and gastric hepatic metastases, although these data were not confirmed in a prospective series including digestive tract and breast liver metastases. Furthermore, low-pretreatment ADC value of colorectal liver metastases has shown an association to shorter overall survival and progression-free survival.

Biliary system and gallbladder malignancies

DWI detects intrahepatic and extrahepatic cholangiocarcinoma (CHC) with advantage over T2-weighted and in a similar manner to DCE-MR imaging. Because this lesion shows the lowest ADC of all hepatic malignancies, its differentiation from benign focal liver lesions using DWI is feasible. DWI also helps in the differentiation between mucus and viable intraductal CHC obstructing the bile duct. The lower the degree of tumor differentiation in CHC, the lower the ADC value is. A target appearance on DWI, a hyperintense peripheral halo (viable tumor) with a hypointense central area (fibrosis), permits the accurate differentiation of small intrahepatic CHC from HCC, because this sign is superior to other morphologic, DCE-MR imaging, or HB phase features.

As with other malignancies, gallbladder carcinoma typically demonstrates high signal on high b-value imaging, and low ADC values ( Fig. 7 ). Acute cholecystitis may simulate a neoplastic process on DWI. For this reason, DWI must be evaluated along with the other morphologic images, increasing the accuracy in the differentiation of benign and malignant gallbladder lesions.

Gallbladder carcinoma with DWI. ( A ) Axial TSE T2-weighted image and ( B ) delayed postcontrast T1 high resolution isotropic volume excitation (THRIVE) show a mural thickening of the fundus of the gallbladder invading the adjacent hepatic parenchyma. Notice the presence of multiple lithiasis. ( C , D ) High b-value DWI with inverted gray-scale and corresponding ADC map demonstrates restriction of water motion of the mural lesion ( arrow ), with low ADC of 1.08 × 10 ⁻³ mm ² /s (ROI).

Pancreatic cancer and other pancreatic tumors

A recent meta-analysis demonstrates high performance of DWI in detection of pancreatic adenocarcinoma (PA), with similar sensitivity to fluorodeoxyglucose (18FDG) PET/CT, but better specificity. Most commonly, PA shows high SI on high b-value imaging and lower ADC values than normal pancreas, related to dense fibrosis and increased cellular elements ( Fig. 8 ). However, PA can have different appearances on diffusion imaging depending on histologic characteristics. Edematous fibrosis and loose collagen fibers have been described in PA to yield higher ADC than normal parenchyma. This heterogeneity in histologic content is probably the cause of contradictory data in the relationship between ADC values of PA and tumor grade. Moreover, the association of acute pancreatitis and PA located in the pancreatic head has been reported. Both entities show a similar behavior in DWI, hampering cancer detection. The use of IVIM-derived parameters, such as f, improves the differentiation between PA and normal pancreatic tissue.

Pancreatic carcinoma ( arrows ) with DWI. ( A ) Axial TSE T2-weighted image shows an ill-defined mass located in the pancreatic head. ( B , C ) High b-value DWI and corresponding ADC map demonstrate restriction of water motion of the lesion, with low ADC value: 0.9 × 10 ⁻³ mm ² /s.

DWI is also helpful in the differentiation between PA and benign lesions, although with limitations in the distinction from mass-forming pancreatitis. There is an evident overlap in ADC values of both lesions, probably due to variable proportions of fibrosis and inflammation in mass-forming pancreatitis and different degrees of PA differentiation (see Fig. 8 ; Fig. 9 ). However, initial data suggest a role for IVIM-derived perfusion parameters in this distinction.

Mass-forming chronic pancreatitis ( arrowheads ). ( A , B ) DCE-MR imaging during the portal and delayed phases show a large mass in the pancreatic head with peripheral enhancement during the portal phase and delayed heterogeneous wash-in. Notice the distal dilatation of main pancreatic duct due to obstruction ( arrow ). ( C , D ) High b-value DWI and corresponding ADC map demonstrate restriction of water motion of the lesion. Notice how the limits of the lesion are better depicted on DWI compared with DCE-MR imaging.

In the therapy monitoring setting, preliminary data suggest that lower pretreatment ADC values of advanced PA, treated with chemotherapy or chemoradiation, are related to poor response and early progression ( Fig. 10 ). DWI is also useful in the evaluation of autoimmune pancreatitis response to steroids. This entity shows high SI on high b value and lower ADC values than chronic pancreatitis and PA. However, with successful treatment, SI on high b value decreases, and ADC returns to normal values.

Therapy monitoring of pancreatic carcinoma ( arrows ) with DWI. ( A ) Pretreatment axial TSE T2-weighted image depicts a large hyperintense lesion involving the distal aspect of the body and tail of the pancreas. ( B , C ) Pretreatment DWI with b value of 1000 s/mm ² and corresponding ADC map show restriction of water motion of the pancreatic tumor with ADC value of 1.4 × 10 ⁻³ mm ² /s. ( D – F ) 5 months after chemotherapy TSE T2-weighted, DWI, and ADC map, respectively, demonstrate stable size of the mass, but with decrease in SI on high b-value image and increase in ADC value: 1.8 × 10 ⁻³ mm ² /s, indicating partial response.

Pancreatic neuroendocrine tumors (PNET) show a restrictive pattern in diffusion, in comparison with normal pancreatic parenchyma ( Fig. 11 ). However, the detection rate of PNETs with MR imaging including DWI was significantly lower than that of ⁶⁸ Ga-DOTATATE PET/CT. Most aggressive PNETs show lower ADC values than benign lesions. In addition, an inverse correlation between Ki-67 (cellular marker for proliferation) and ADC values has been established, which would allow for the evaluation of tumor aggressiveness. Perfusion parameters from IVIM models are able to accurately differentiate PNET from PA.

PNET. ( A ) Axial TSE T2-weighted and ( B ) postcontrast e-THRIVE during the arterial phase demonstrate a large mass in pancreatic head with vessel involvement. The mass shows high SI on T2-weighted image and intense heterogeneous enhancement. ( C , D ) DWI with b value of 800 s/mm ² ( arrow ) and corresponding ADC map confirm the aggressiveness of the lesion with intense restriction of water motion (ADC: 1 × 10 ⁻³ mm ² /s).

Pancreatic cystic lesions are a common incidental finding. It is important to differentiate cystic tumors from nonneoplastic lesions and identify malignant variants. Cystic tumors show different biological behavior and risk of malignancy and can be classified as shown in Box 3 . Cystic lesions greater than 2 cm in size are usually premalignant or malignant.

DWI with high b value can differentiate nonneoplastic cysts, such as simple cysts and pseudocysts, which are usually isointense to the pancreas, from neoplastic cysts and abscesses, which remain hyperintense. Boraschi and colleagues found statistically significant differences in ADC values of intraductal papillary mucinous neoplasms (IPMNs), mucinous cystic neoplasms (MCNs), serous cystadenomas, and pseudocysts, although other series have shown significant overlap in the ADC values of cystic pancreatic lesions. More consistent data have shown significantly higher ADC values for IPMNs compared with MCNs. Furthermore, ADC and D values of malignant IPMN are significantly lower than benign variants ( Figs. 12 and 13 ). Currently, the differentiation of cystic pancreatic lesions only based on DWI remains limited, although integrated in a comprehensive MR protocol, diffusion is considered helpful.

Accessory-branch IPMN. ( A , B ) Axial TSE T2-weighted and coronal thick-slice 2-dimensional cholangiography demonstrate a cystic mass with small locules and septa in the pancreatic head. The mass is connected to a secondary branch of the pancreatic duct. ( C , D ) DWI with b value of 800 s/mm ² and corresponding ADC map show high signal on high b-value image ( arrow ), but absence of true restriction of water motion of the pancreatic tumor with ADC value of 2.9 × 10 ⁻³ mm ² /s (ROI). This appearance is due to T2 shine-through effect.

Malignant MCN. ( A , B ) Axial TSE T2-weighted image and ( B ) axial postcontrast THRIVE during the venous phase show a complex cystic lesion with solid peripheral neoplastic component that shows enhancement and a mural nodule ( arrows ). ( C , D ) Diffusion-weighted image ( b : 800 mm ² /s) with inverted gray-scale and corresponding ADC map demonstrate restriction of water motion in the solid mural thickening ( arrowhead ), with ADC value of 1.04 × 10 ⁻³ mm ² /s. ADC of the cystic component is 2.4 × 10 ⁻³ mm ² /s (ROIs).

Image acquisition

The characteristics of an ideal DCE-MR imaging perfusion examination are listed in Box 4 . Gadolinium-based contrast agents (GBCAs) are used as tracers and image acquisition aims to accurately capture changes in tissue SI, which is related to changes in concentration of the GBCA over time. Both extracellular and hepatobiliary contrast agents (HBCA; Gd-EOB-DTPA) have been used for perfusion MR imaging studies. Before injecting the contrast agent, a T ₁ quantification acquisition is performed to determine the baseline tissue signal. This T ₁ quantification acquisition is done by using the variable flip angle technique ; flip angles in the range of 30° to 60° are recommended. This T ₁ quantification acquisition is followed by the actual DCE-MR imaging acquisition: a GBCA (0.1 mmol/kg body weight) is injected through a wide-bore cannula (20 G or larger) placed in a large antecubital vein using a power injector at a flow rate of 2 to 4 mL/s, followed by a 20 to 30 mL saline flush. Repeated T ₁ -weighted gradient echo imaging of the tissue of interest is performed every few seconds. The capillary transit time in well-perfused organs such as the liver is on the order of 3 to 5 seconds; hence, a sampling interval of less than 2 seconds per volume is typically required. Multiple volumes are acquired over a period of 3 to 5 minutes, starting 10 to 20 seconds before contrast injection, to capture the change in concentration of the GBCA with time.

Box 4

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

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related posts:

Body Diffusion-weighted MR Imaging in Oncology Assessment of Tumor Angiogenesis Clinical Imaging of Tumor Metabolism with 1H Magnetic Resonance Spectroscopy Multiparametric MR Imaging of Breast Cancer Role of Multiparametric MR Imaging in Malignancies of the Urogenital Tract Functional MR Imaging in Gynecologic Cancer

Stay updated, free articles. Join our Telegram channel

Join