The potential of PET/MR for brain imaging

The progress in image quality may be followed by metabolic images of glucose consumption in one volunteer acquired on several generation PET systems over the years. The first transaxial image (Fig. 1a) was acquired on a single hexagonal array, the ECAT PET, which covered only the cortical rim. Early PET images suffered from limited spatial resolution (approximately 15 mm), low sensitivity and insufficient attenuation and scatter correction [3, 4]. PET image quality improved with the four-ring PC-384 (Scanditronix) [5], which yielded seven simultaneous, partly overlapping transaxial images and that provided a spatial resolution of 8.4 mm FWHM across 12-mm slices (Fig. 1b) [6] and limited axial reconstruction (Fig. 2). Further improvements came with the ECAT EXACT PET, which yielded 47 contiguous image planes with a spatial resolution of 6.5–7 mm and 5–8 mm in the transverse and axial directions, respectively (Fig. 2) [7]. For the first time the entire brain could be imaged in a single PET examination. The next generation PET system, the ECAT EXACT HR, provided further improvements of the spatial resolution (3.6 mm and 4.0 mm in the transverse and axial direction, respectively), sensitivity and image quality [8] (Figs. 1 and 2). With the dedicated high-resolution research tomograph (HRRT) PET image resolution improved to 2.3 mm and 2.5 mm in the transaxial and axial directions, respectively, while sensitivity has increased to 4.3% [9, 10] (Figs. 1 and 2).

The most attractive feature of PET is the high sensitivity and specificity in detecting biochemical and molecular tracers, but it suffers from limited spatial resolution. Therefore, coregistration of PET images with morphological imaging modalities such as CT and MRI has been studied [11, 12], but is not yet in routine clinical use. Research applications of image fusion software have also been challenged by, for example, mapping of molecular and genetic activities to certain cortical areas.

The integration of PET and CT into one imaging system permitting a simple overlay of sequentially acquired images has been a great success in clinical diagnosis [13]. However, because of the limited soft-tissue specificity of CT, this integrated system is not of great value for brain imaging. The installation of a HRRT PET system adjacent to a high-field 7T MRI system allows the utilization of high sensitivity molecular/biochemical and high-resolution anatomical imaging for brain research [14]. However, this system is not integrated and therefore does not permit simultaneous PET and MR data acquisition. For that purpose an integrated PET/MR system is required, which would permit the simultaneous coregistration of various MR and PET procedures. Schmand et al. demonstrated for the first time that simultaneous PET and MR data acquisition is feasible with an integrated PET/MR tomograph [15]. Figure 3 shows the first simultaneously acquired MR and PET images [16] in a volunteer and demonstrates the feasibility of combined MR and PET data acquisition without significant interference between the two modalities. Conventional MR imaging on a 3-T MR tomograph was performed using a T2-weighted turbo-spin echo sequence, a fluid attenuated inversion recovery sequence and a 3-D T1-weighted fast low-angle shot sequence. Additionally, diffusion-weighted imaging, time-of-flight MR angiography (MRA) and proton MR spectroscopy were performed during the acquisition of the PET data. For PET imaging 370 MBq of ¹⁸F-fluorodeoxy-d-glucose was intravenously injected and the activity in the head was accumulated over 40 min, starting 20 min after tracer injection. PET images were reconstructed using a 3D-OSEM algorithm following scatter, randoms and attenuation correction. The resolution of the reconstructed images was 2.5 mm FWHM at the centre and 4.5 mm at 10 cm off-axis with a sensitivity of 5.6%. Overall, the quality of the images was comparable to that of the corresponding stand-alone PET and MRI systems [16].

With the installation of PET/MR systems in a clinical environment, this integrated imaging modality, which is already utilized in preclinical research already is on the verge of being applied to the clinical neurosciences [17, 18]. An integrated PET/MRI system will permit the simultaneous acquisition of several parameters (Table 1). Quantitative values from PET (some in the picomolar range) of a large number of biological parameters are complemented by the high-resolution information provided by MRI (in the micromolar range) to yield complementary information of previously unexpected variability.

The combination of imaging modalities for high sensitivity and high resolution with the additional advantage of utilizing dynamic acquisition procedures appears very appealing for a variety of clinical and research applications. Recently, PET data and MRI data have been retrospectively combined for detection and staging of gliomas [19, 20] as well as for identification of areas with critical neurofunction in the vicinity of tumours, which is important for planning surgery [21]. Image fusion has gained a place in the early diagnosis of dementia and mild cognitive impairment [22] and degenerative disorders, e.g. cerebral atrophy and Huntington chorea. Image fusion is of clinical value in the detection of epileptic foci accessible for surgery [23] and for the identification of metabolic activity, transmitter concentration and enzyme expression in small brain structures [24]. In experimental focal brain ischaemia and ischaemic stroke, coregistration of regional values for cerebral blood flow, oxygen utilization and glucose metabolism on early and late MRI has permitted the differentiation of irreversibly damaged areas and functionally impaired but morphologically preserved areas, thus affecting treatment strategies [25].

The established applications of coregistered MRI and PET will gain new dimensions with the simultaneous procedures made possible by integrated PET/MR. Simultaneous data acquisition will allow the addition of kinetic, functional and metabolic information for real-time multiparametric imaging. The vascular origin of changes in cerebral blood flow, oxygen consumption and metabolism causing stroke and intracerebral and subarachnoid haemorrhages can be detected by MRA. Perfusion changes can be related directly to the extent of hypoxia (¹⁸F-MISO-PET) and changes in metabolic markers (e.g. lactate, choline and N-acetyl-aspartate) can be assessed through simultaneous PW-MRI, MRS and ¹⁸F-MISO-PET. Diffusion- and perfusion-weighted (DW-PW) MRI can be performed during PET measurements of ¹⁵O and H₂ ¹⁵O for differentiation of intervascular perfusion, tissue blood flow, penumbra and irreversible tissue damage in ischaemic stroke [26]. This differentiation is important for therapeutic decision. For that purpose the validation of MR surrogate measures (mismatch between PW- and DW-MRI) is needed. The combination of various MRS procedures with PET will broaden the insight into the complex metabolic changes caused by brain diseases.

Coregistration of image data from the different modalities has indicated the value of such comparative studies in stroke [27], gliomas [28], and degenerative [29] and congenital or hereditary disorders [30]. However, coregistration was limited since it required multiple investigations in different laboratories. Real-time simultaneous studies will be of particular interest for the interpretation of activation patterns (intravascular versus tissue signals) obtained with fMRI and H₂ ¹⁵O (or FDG) PET [31, 32] and will be useful for the analysis of the effect of specific tasks on transmitter release and receptor binding, such as the determination of DOPA, raclopride and 5-hydroxy-tryptamine transporter by combining fMRI with PET [33‐35]. A similar approach (fMRI and PET of receptor ligands) might be useful to elucidate the effects of drugs and their withdrawal (e.g. nicotine) on task performance. However, in these combined applications of fMRI and PET the differences in temporal resolution and sampling must be considered: separated short time frame MRI (~ seconds) versus PET acquisitions of several minutes.

The high temporal resolution of MRI can be utilized for the acquisition of dynamic data can be utilized for the quantification of metabolic values by PET. For example, flow-dependent kinetic constants can be determined by PWI and used for compartmental analysis of PET data (e.g. FDG, FLT) [36] or for the assessment of the kinetics of the distribution of tracers or labelled drugs in various brain structures [37].

Simultaneous data acquisition by MRI and PET could also be used to develop new methods, e.g. values of cerebral blood flow obtained by arterial spin labelling MRI can be validated by H₂ ¹⁵O-PET [38], and ¹⁷O as a new MRI tracer for oxygen utilization [39] can be compared to cerebral metabolic rate of O₂ as determined by PET. This promising new MRI method uses inhalation of air enriched with ¹⁷O₂ whereby the inhaled ¹⁷O is converted into H₂ ¹⁷O in proportion to the oxygen consumption. The metabolite H₂ ¹⁷O perturbs the proton signal resulting in negative contrast on T2-weighted MRI. Thus, the signal intensity is proportional to oxygen consumption.

For complex studies of brain function special combinations of modalities or additional investigative procedures may open new insights into the organization of the brain and diseases–related changes of brain function. For these studies diffusion tensor tracking will add new dimensions, since it could be performed in close temporal relation to activation studies mapping the effects on transmitter release and receptor occupancy as well as on metabolism in connected areas of functional networks [40]. With this approach the effect of deep brain stimulation, which is a means to ameliorate neurological symptoms and to improve abnormal behaviour [41] could be demonstrated. The effect of such an invasive manipulation on metabolism in defined regions together with the connecting fibre tracts is shown in Fig. 4. It may also be possible to analyse the effect of repetitive transcranial magnetic stimulation (rTMS), which can be used to activate or inhibit selected areas of the cortex, on regional metabolism, and on the involvement of transmitter/receptor systems and the connecting network [42].

Fusion of PET and MR images has been used in several attempts to identify molecular events induced by local pathogenetic factors, by the transfer of genetic material or implantation of (stem) cells. While the above studies were performed using separate imaging modalities, it is possible that these concepts could be transferred more easily to routine applications in humans through an integrated PET/MR tomograph, by which various functional and morphological aspects can be observed simultaneously. In several of these applications the identification of small regions or cell clusters by MRI might be essential for the analysis of biochemical processes by PET. Therefore, these examples may create new combined applications of MR and PET.

Angiogenesis is a fundamental process in various physiological and pathological processes. Visualization, quantification and monitoring of angiogenesis is of interest in various fields, including oncology and cardiology. Molecules regulating angiogenesis include growth factor receptors and integrins. Cyclic RGD peptides bind to integrin α_νβ_3, and ¹⁸F-galacto-RGD has been tested in humans with high uptake in highly vascularized tumours. Alternatively, labelled ligands have been used to study the vascular endothelial growth factor receptor. When combined with dynamic, contrast-enhanced MR, yielding vascular and extravascular volumes, vascular permeability and perfusion angiogenesis PET markers may permit the follow-up of angiogenetic and antiangiogenetic treatment [43].

Targeted gene transfer by various vectors can be used to express foreign enzymes in cells. This strategy can be applied to make malignant cells susceptible to specific drugs, which are toxic only to those cells expressing the enzymes. In an experimental glioma model the efficacy of this treatment strategy with transfer of the herpes virus thymidine kinase gene and application of ganciclovir has been demonstrated and preliminary clinical tests have shown some efficacy of this treatment [44]. The effect of this therapy can be followed by multitracer PET and verified on MRI (Fig. 5).

Implantation of embryonic stem cells into the striatum lesioned by injection of 6-OHDA in rats leads to proliferation and differentiation of dopaminergic cells. This was demonstrated by coregistering MRI and PET of the specific dopamine transporter ligands ¹¹C-CFT [¹¹C-2β-carbomethoxy-3β-(4-fluorophenyl) tropane] where CFT binding was restored in the region of TH-immunoreactive neurons documented post mortem. The restored functional activity of these implanted cells was demonstrated by the response to amphetamine, which caused an increase in rCBV due to dopamine release [45]. Simultaneous PET/MR can be used to show the viability and differentiation of transplanted cells and their effect on neuronal networks [46]

Cell replacement approaches are an innovative strategy for treatment of various neurological disorders. For the development of these approaches in animal models it is essential to monitor the location and to follow the migration of grafted stem or progenitor cells. Various strategies have been proposed for labelling these cells with ultrasmall superparamagnetic iron oxide particles and micron-sized iron oxide particles [47]. In focal experimental ischaemia labelled stem cells migrated over 3 weeks from the contralateral implantation site along the corpus callosum to the ventricular walls and massively populated the border zone of the damaged brain tissue on the lesioned hemisphere (Fig. 6). A combination of MRI for tracking cells and of PET for proving their biological activity could demonstrate the viability of the cells as well as their integration into functional networks [48]. Monitoring cell viability and migration by MRI combined with indicators of function from PET might also become a qualifying step in strategies relying on transplantation of fetal grafts, e.g. in Parkinson disease.