Carotid Angiography

In terms of plaque visualization and segmentation, although CT is able to achieve submillimeter spatial resolution in combination with a very short acquisition time, soft tissue contrast is poor and 1-dimensional (ie, plaque segmentation is based purely on photon attenuation). CT carotid imaging is also hampered by blooming artifact from dense calcification. Both these factors make visualization of the luminal and adventitial borders for region of interest analysis difficult. By comparison, MR with black-blood imaging offers superior soft tissue detail that provides excellent contrast at the adventitial and luminal borders (see Fig. 1 ). Iodinated contrast enhancement does improve visualization on CT but may be contraindicated in renal failure or with previous contrast reactions and does nothing to abrogate blooming from calcium. Although MR angiography with gadolinium remains a concern in patients with advanced renal failure, alternative technique can be performed safely in these patients either via the use of alternative contrast agents (iron oxide nanoparticles ) or using standard time of flight (TOF) MR sequences.

Limitations of Histologic Validation and Cross Validation with MR Imaging

A standard and productive model for cardiovascular PET research has been to scan patients just before carotid endarterectomy (CEA). The excised plaque can then be collected and used to validate the PET radiotracer in question using histologic or even gene expression techniques. This model is, however, becoming difficult to deliver as a result of the increasing impetus to intervene expeditiously on patients with symptomatic carotid stenosis and a greater reluctance to intervene on asymptomatic lesions. A further consideration is that the role of carotid artery stenting may increase, reducing the role of endarterectomy and the opportunity to use this valuable approach. Histologic examination also has its limitations: patients undergoing endarterectomy are invariably at the severe end of the disease spectrum, biasing such studies against the earlier phases of the condition. Moreover, the processing of the heterogeneous and calcified necrotic tissue that emerges from surgery is notoriously difficult, and further complicated by trauma to the specimen at the time of its excision. Finally, histologic sampling is necessarily limited to small regions of the plaque and thus prone to sampling bias. Critically, these regions do not include the outer tunica media and tunica adventitia of the carotid artery (because of the way the plaque is excised), tissues that are increasing recognized to play central roles in atherogenesis.

Alternative methods of validating novel PET radiotracers and imaging techniques are therefore required. Because of the limitations previously discussed, CT is not able to adequately resolve carotid plaque for detailed compositional and phenotypic analysis. By contrast, MR provides much greater information with respect to plaque morphology, with the ability to identify multiple key plaque characteristics, including an intact or ruptured fibrous cap, a necrotic lipid-rich core, intraplaque hemorrhage, thrombus, intraplaque neovascularization, macrophage infiltration, biomechanical stress, and others ( Fig. 2 ) (a recent review by Makris and colleagues is recommended ). Although obviously not as “definitive” as histology, MR is an excellent alternative and complementary technique with the key advantage that plaques can be assessed in situ at multiple stages in their development. MR also permits serial and 3-dimensional volumetric assessment of the carotid artery in its entirety, including the layers missed with histologic validation. On this basis, hybrid PET/MR offers the opportunity to cross validate a wide range of novel PET tracers, providing mechanistic information beyond simple anatomic coregistration ( Table 2 ).

Example of atheromatous carotid plaque on multicontrast MR (3T) and [18F]-fluoride PET/CT. Images are all from the same left proximal internal carotid artery (ICA) in a neurologically symptomatic patient. The red arrow denotes the lumen of the ICA and the yellow star denotes the hemorrhagic plaque. ( A ) Axial TOF MR angiogram. ( B ) T2-weighted image. ( C ) T1-weighted image. ( D ) CT carotid angiogram. ( E ) Fused PET/CT. ( F ) [18F]-fluoride PET.

( Data from authors’ institution.)

Indeed, the feasibility of this approach has already been demonstrated in several clinical studies using sequentially acquired [18F]-FDG PET/CT and MR, studies that would have been greatly simplified by a hybrid system. Moreover, 2 elegant preclinical studies have recently been published in this domain, the first cross-validating a fibrin-targeted probe (EP-2104R) and the second comparing ultrasmall superparamagnetic iron oxide–enhanced MR with [18F]-FDG as alternative markers of plaque inflammation. The availability of such preclinical PET/MR systems, capable of scanning human specimens ex vivo and small animals in vivo, is likely to accelerate the development and clinical translation of this imaging technique.

Simultaneous Scanning and Motion Correction

CT and PET datasets are necessarily acquired sequentially and are frequently either degraded if not frankly spoiled by movement. The nature of this movement can either be intrinsic (ie, cardiac motion, respiratory motion, swallowing, hollow viscus peristalsis) or extrinsic (ie, an uncomfortable patient who squirms or shifts position). The ability to simultaneously acquire PET and MR data is a critical advantage that modern hybrid PET/MR systems hold. This opens up the possibility of accurate motion correction with huge potential benefits for imaging of the coronary arteries and the myocardium in addition to the carotids ( Fig. 3 ).

The potential for motion correction using PET/MR. Each image is the same [18F]-FDG PET slice (with simulated “hot plaque”) reconstructed with and without motion correction using a method based on fat segmentation. A variety of target-to-blood activity ratios are demonstrated. Motion correction can be clearly seen to significantly improve the ability to resolve the increased intraplaque uptake ( yellow arrows ).

( Data from Petibon Y, Fakhri El G, Nezafat R, et al. Towards coronary plaque imaging using simultaneous PET-MR: a simulation study. Phys Med Biol 2014;59(5):1203–22; with permission.)

With PET/CT, the only options for tackling motion relate either to measures aimed at controlling such movement (ie, asking patients to lie still, ensuring their comfort, or using neck collars and blocks to provide gentle restraint) or to postprocessing the PET data in an attempt to correct for it. Postprocessing may involve either registration (manual or automated, rigid or nonrigid) that is limited by poor PET spatial resolution (∼4.5–5 mm) and a lack of fiducial landmarks or, in the case of cardiac and respiratory motion, retrospective gating. Retrospective gating is accomplished by acquiring PET data in list mode (analogous to using a video camera instead of a photographic camera) and simultaneously acquiring electrocardiogram and respiratory cycle data. This permits retrospective exclusion of PET data acquired during portions of the cardiac and/or respiratory cycle where the region of interest is unacceptably mobile. Unfortunately, discarding most PET events with retrospective gating has a major negative impact on the signal-to-noise ratio (SNR) that can presently only be abrogated by increasing radiotracer dose or patient bed-times. Additionally, this retrospective approach does not typically attempt to correct for the effects of patient motion when constructing the attenuation correction maps, leading to the potential for systematic biases in PET quantification.

Hybrid PET/MR can permit improved retrospective gating either by directly measuring organ location (eg, by using a diaphragmatic navigator) or far more elegantly, by obtaining continuous and synchronous MR and PET data. Indeed, the latter approach offers the theoretic and enticing possibility of correcting for both intrinsic and extrinsic movement without having to discard any emission data. Not only would this significantly enhance image quality by minimizing motion blur and improving SNR, it would also have major cost implications, as scanning times could be reduced allowing more patients to be scanned in a given scanning session or production run on the cyclotron. Ultimately, such economic factors could prove crucial in translating cardiovascular PET imaging into the clinical environment.

Robust motion correction using PET/MR might also make accurate dynamic PET imaging of moving tissues a possibility. The general principle of MR-based motion correction involves recording the displacements of the relevant anatomy in space and time (ie, in 4 dimensions) using MR and then using these data to either retrospectively “deconvolve” the reconstructed and time-binned PET data, or better still, to achieve this prospectively by including the 4-dimensional MR data in the reconstruction algorithm. In principle, the same methods also may be used to reduce the impact of movement on attenuation correction, thus further improving data quality. It is beyond the scope of this review to discuss these methods in detail but the reader is referred to excellent articles by Catana and Petibon and colleagues. Although the biggest impact of these advances may well apply to PET imaging of the heart, such motion correction also will be of direct relevance to carotid imaging. Indeed, little can presently be done to control for small extrinsic movements or the effects of swallowing and jaw movement, which can have a major impact on PET quantification in small-volume carotid plaques.

Radiation Exposure, Serial Imaging, and Multiprocess Imaging

One of the major attractions of PET is that it is able to resolve specific biological pathways and physiologic activity in vivo with exquisite sensitivity and without disturbing the subjects’ biology. This had led to a new model for phase 2 clinical trials in which novel pharmacologic agents are tested for biological efficacy in vivo using serial PET examinations, before proceeding to large and expensive phase 3 trials. Indeed, we successfully used this approach in the recent dal-PLAQUE study and are currently conducting several other similar studies in our institutions. One of the major concerns relating to this approach is the large cumulative dose of radiation that may accrue with multiple PET/CT examinations. The ability to remove CT from the equation will go a long way to allaying anxiety on this front, and in opening up the possibility of examining disease activity at 3 or 4 different time points, or investigating several disease processes simultaneously.

Potential for Multiprocess Imaging and PET Sensitivity and Specificity

As well as permitting cross validation, PET/MR offers the opportunity to combine the functional information of PET with emerging molecular MR techniques, rendering true multimodality, multiprocess imaging possible. This would allow for studies, incorporating a wide array of both PET and MR techniques, capable of investigating multiple pathologic processes simultaneously. A complete discussion of emerging molecular MR techniques is beyond the scope of this review, but includes the design of specific molecular MR tracers incorporating either paramagnetic (gadolinium) or superparamagnetic (coated iron oxide nanoparticles) reporters (see Table 2 ), dual-tuned coils to enable simultaneous [1H] and [19F] imaging of perfluorocarbon tracers, MR spectroscopy, and hyperpolarized MR (indeed, the concept of “HyperPET” has already entered the lexicon ). The opportunity also exists to design imaging nanoparticles with reporters for both PET and MR. Although promising, several of these MR techniques are currently limited either by the nonlinearity of reporter accumulation and signal change or by low sensitivity. Indeed, PET is generally quoted to be 3 to 4 orders of magnitude more sensitive than MR, which of course is a key attraction of PET.

Versatility and Concurrent Imaging of Other Relevant Organ Systems

MR imaging is currently the gold standard technique for imaging the brain, and many of the radiotracers of interest in the carotid also will be of interest in the brain (eg, [18F]-FDG). The ability to do a full simultaneous vasculo-cerebral PET/MR assessment using multiple complementary sequences (eg, diffusion-weighted imaging abnormalities or cerebral perfusion imaging in the context of a carotid stenosis) and PET acquisitions will therefore be of major interest to stroke and vasculopathy researchers, especially as it will not involve additional radiation exposure.