Simultaneous dual-contrast multi-phase liver imaging using spectral photon-counting computed tomography: a proof-of-concept study

In this study we present our first results of an in-silico simulation of simultaneous dual-contrast multi-phase SPCCT liver imaging. A dedicated perfusion protocol was defined to simultaneously assess the arterial and portal venous distribution of two CAs. We evaluated the typical presentation of the arterial and portal venous contrast enhancement for four characteristic types of liver lesions. Further, the likelihood maps, based on the information of both CAs in different dynamic phases, illustrated areas with high probability for the respective lesion within normal liver tissue. This new dual-contrast SPCCT protocol therefore shows a potential for a clinically relevant simultaneous assessment of the arterial and portal venous contrast enhancement of liver lesions at one time point using a single SPCCT scan acquisition, together with a virtually unenhanced image. This approach goes beyond the currently possible ability of dual-energy systems, which can be used to acquire simultaneous iodine and gadolinium images [12], but not a third additional virtual unenhanced image simultaneously.

In SPCCT, the x-ray photon is directly converted into an electrical pulse signal, whose amplitude is indicative for the energy of the initial x-ray photon [13–17]. Without considering K-edge absorption, the energy-dependent attenuation of the human body can be modelled as a linear combination of a set of materials [18] and is then typically expressed in a basis decomposition of two materials or, in terms of the underlying physical interaction processes, photo-electric effect and Compton-scattering. If K-edges are taken into account, some materials feature an abrupt increase of the attenuation due to resonant ionisation of a K-shell electron at a certain energy (K-edge) within the energy range relevant for imaging, roughly from 40 keV up to 120 keV. This characteristic attenuation behaviour at the K-edge can be used to perform a specific material decomposition with a high signal-to-noise ratio. Gadolinium features a K-edge at 50.24 keV and thus is an ideal candidate for K-edge CT imaging, even better than iodine, which is the standard CA for contrast-enhanced CT scanning with a relatively low K-edge energy at 33.17 keV. SPCCT allows for a valid differentiation between iodine and gadolinium, even if both iodine-based and gadolinium-based contrast CAs can have the same density values on standard CT [10].

In SPCCT, material decomposition delivers a set of basis images, which typically features higher noise than an integrated image. But the noise in the decomposed images is anti-correlated so that, using recent reconstruction methods, this noise can efficiently be reduced [19]. Such de-noising methods not only lower the noise, but also yield a higher contrast resolution.

Notably, liver parenchyma receives a dual blood supply from the portal vein and hepatic artery. Normal liver parenchyma is mainly perfused via the portal vein, whereas HCCs are mostly supplied with blood from the hepatic artery. Therefore, information of perfusion patterns has an important impact on diagnostic decision-making [20]. Thus, a dedicated liver protocol includes at least one arterial and one portal venous contrast phases. Individual criteria can also require a native unenhanced scan, e.g. for the detection of calcifications and in patients with a history of transarterial chemoembolisation with iodined oil [21]. According to the Dose Index Registry Report from the American College of Radiology, the median radiation dose in terms of dose-length product for a standard single CT acquisition of the abdomen is 445 mGy × cm [22]. As a consequence, triple scan examination of the liver leads to a radiation dose of around 1335 mGy × cm. Dual-contrast SPCCT of the liver offers detailed three-phase contrast information with radiation exposure similar to that needed for a single CT scan of the abdomen.

With standard single-energy CT, a differentiation between normal liver parenchyma and focal liver lesions is achieved by comparing the changes in contrast enhancement between arterial and portal venous phase. In the past, poorly aligned multi-phase CT examination protocols made it difficult to assess these changes in contrast enhancement of the same anatomic areas and post-processing techniques (including motion-correction of repetitive CT scans) were essential for reliable results of extracted perfusion parameters [23, 24]. Using the method presented here, all dynamic information is extracted from a single SPCCT scan and hence is inherently co-registered (on a voxel-by-voxel basis), circumventing any previous issues due to subsequent multi-phase scans or motion artifacts (e.g. those determined by different depths of breathing). This further allows for a pixel-by-pixel evaluation of contrast enhancement of a lesion, as advanced analysis using scatter plots and likelihood maps require a perfect spatial match of the vector space components. This perfect match can be provided by the protocol here proposed.

Of note, we have used a gadolinium-based agent as first CA and an iodine-based agent as second CA. In future studies it will be necessary to address the question if this order is the preferred one or if iodine-based agent as first CA and gadolinium-based agent as second CA might be advantageous.

One future perspective of our approach could be adding a delayed dynamic phase, a clinically relevant issue particularly for characterising haemangioma and HCC [25]. While technologically feasible, such an approach would require a third CA. This, however, is presently a challenge, as iodine-based CAs and gadolinium-based CAs are the only available CAs with a potential for this use.

There are limitations of our study. First, we used numerical experiments to evaluate the imaging features in dual-contrast liver SPCCT. Although the simulations were performed using sophisticated tools and incorporated most physical effects in the SPCCT system, the findings have to be proved in an in-vivo study with a real SPCCT system in future. Second, there is currently very little clinical experience with the subsequent injection of iodine-based CA and gadolinium-based CA in humans. Therefore, pharmacological testing of potential adverse side effects will be necessary before an application of our presented contrast injection protocol can be used in patients.

Third, there is a general limitation with respect to the minimum lesion size assessable with this approach, as the contrast uptake and characteristic patterns cannot be studied when the lesion becomes too small. Finally, we should consider pharmacodynamic considerations about the long-term deposition of gadolinium in patients, which presently is a subject to intense discussions in the community.

In conclusion, we proved the concept of single-scan dual-CA CT as a new approach for K-edge SPCCT imaging, which exploits the spectral information for simultaneously assessing the gadolinium-based and iodine-based liver enhancement in different dynamic phases. Our simulation results have shown that we can successfully detect and visually discriminate the clinical presentation of four typical liver lesions by simultaneously evaluating unenhanced, arterial, and portal venous phases at one time point by a single scan, with a potential for dose reduction in clinical practice.