A PET scanner images the distribution of radioactive tracer agents that have been injected into the body. The common PET tracer 2-fluoro-2-deoxyglucose (FDG), for example, is an analogue of glucose and therefore accumulates in metabolic hotspots like rapidly proliferating cancer cells. This allows primary tumours and their metastases to be seen on a PET image.

In step with the latest developments in molecular medicine, PET techniques have been developed for imaging the delivery of molecularly targeted drugs, the delivery and expression of therapeutic genes for gene therapy, and the distribution of stem cells for cellular therapeutics. This has made PET a widely used modality for clinical diagnostics, clinical research and also increasingly for preclinical research.

The trouble is, although PET scanners can readily detect the faint signals from the decay of the positron-emitting radionuclides used as tracer agents, it is often difficult to pinpoint the precise location of those signals within the body. Only cells and tissues that accumulate the radioactive probe are visible in a PET scan, and therefore very little anatomical information is available.

Add to this the relatively poor spatial resolution achievable with PET scanning (2 mm at best in humans, 0.5 mm at best in rodent work) and it becomes hard to know exactly where the signal originates. Think of a room containing objects that glow in the dark: the PET scan corresponds to viewing the room with the lights out. The objects can be seen, but to know their precise location in the room, we need to turn the lights on.

Better together

Enter imaging techniques like X-ray CT and MRI, which excel at providing anatomical information. By combining PET with one of these modalities you can have the best of both worlds: sensitive detection of specific molecular targets with PET superimposed on high-resolution anatomical information provided by CT or MRI.

To some extent, this can be accomplished by using a computer to register images acquired on two separate scanners. However, software registration is often not straightforward outside of the brain because tissues can change location and shape based on exactly how the patient lies on the scanner bed. There are also issues related to the movement of stomach and intestinal contents and the filling or emptying of the urinary bladder during scanning. A combined scanner can solve these problems and also make multimodality studies more practical to perform.

Scanners that combine PET with CT have been around for about a decade, and were rapidly adopted for routine clinical diagnostics. In fact, in recent years more than 90% of all PET scanners purchased have been combined PET/CT systems. This clearly demonstrates the dominance of this technology and the value of integrated molecular/anatomical imaging.

Although the PET/CT concept was revolutionary from a technological perspective, it was a relatively simple combination to realize. PET and CT scanners were placed back-to-back in a tandem configuration and integrated primarily through software, with very few hardware modifications required. With such a system, the PET and CT scans are acquired sequentially, via a common patient bed that moves through the two scanners.

Given the success of PET/CT scanners, why is there any need for a combined PET/MRI scanner? One reason is that there are clinical applications where MRI is the preferred anatomical imaging modality, such as pelvic and musculoskeletal imaging. More important, however, is MRI's tremendous flexibility for imaging. The mechanism of signal contrast allows it to reach beyond anatomy and address a much broader range of questions compared with CT imaging.

Areas where MRI has proven useful include looking at changes in blood oxygenation upon brain activation (functional MRI), using diffusion-tensor imaging for neuronal tract tracing, contrast-enhanced MRI in vascular imaging for delineating regions of enhanced permeability, and labelling cells for cell trafficking studies. And then magnetic resonance spectroscopic imaging (MRSI) adds a whole host of other applications to this list.

It is these additional capabilities that excite us in the context of combined PET/MRI systems. The ability to correlate multiple physiological, metabolic or molecular signals though PET and MRI, accurately coregistered with the high-resolution anatomical localization provided by MRI, offers some tremendous opportunities. It will improve the way we use imaging to study complex biological systems in vivo, and there is the possibility of ultimately being taken into the clinic for approaches that can offer demonstrable benefits for disease diagnosis and patient care.

Technically impossible?

Unlike combined PET/CT scanners, however, the practical and technological hurdles that must be surmounted in a PET/MRI system are daunting. Many of the applications we envisage involve dynamic PET and MRI and/or fairly lengthy imaging times, so are not consistent with a scanner that does sequential PET and MRI.

At the University of California Davis, we therefore pursued a different solution and have been developing a PET scanner within a MRI scanner - an arrangement that will permit imaging of the same tissue volume with both modalities simultaneously. Admittedly, this approach is more technologically challenging because of the many potential sources of interference between MRI and PET. These sources of interference stem from such fundamental aspects as the high static magnetic field in the MRI scanner; the rapid generation of magnetic-field gradients and high-intensity RF pulses required to create MR images; and the PET detector materials and electronics placed in the magnet. In fact, it is fair to say that a few years ago many people thought that simultaneous PET/MRI was actually impossible.

Yet recent data from a number of research groups now clearly indicate that it can be done. The use of scintillator materials that have appropriate magnetic susceptibility, combined with magnetic-field-insensitive avalanche photodiodes, allows the photons emitted from the decay of a positron-emitting radionuclide to be detected in the magnet. And with careful shielding of the PET detector preamplifier electronics, electromagnetic interference between the PET and MRI scanner components can be reduced to levels where it is very hard to detect any performance degradation versus independent operation of the two devices.

While many refinements can be anticipated, and further development and validation is clearly needed, first-generation PET/MRI scanners are now available for initial research studies in humans and animal models. The first in vivo studies with simultaneous PET/MRI have now successfully been performed by the team here at UC Davis and several other institutions. Industry is showing an interest as well, with Siemens Medical Solutions recently unveiling a PET/MRI scanner for human brain studies.

The future of PET/MRI scanners hinges on whether this technology can be used to perform studies that were previously impossible and thereby generate new scientific knowledge. Equally, whether clinical applications arise from these studies that demonstrate that PET and MRI need to be performed at the same time. The jury is still out on the question of whether PET/MRI will become a mainstream imaging technology, but the potential is clearly there and we are in for an exciting few years as these issues get explored in more depth.