Stefan Vollmar

The ECAT HRRT: performance and first clinical application of the new high resolution research tomograph

The ECAT HRRT is a three-dimensional (3-D) only dedicated brain tomograph employing the new scintillator lutetium-oxy-orthosilicate (LSO) and using depth of interaction (DOI) information to achieve uniform isotropic resolution across a 20-cm diameter volume. With its unique technological innovations it represents the prototype of a new generation of high-resolution brain tomographs. The physical performance with respect to count rate, live time, scatter, sensitivity, and resolution was evaluated with phantom studies and measurements with a point source. The HRRTs imaging performance was tested with phantoms and fluorodeoxyglucose (FDG) scans performed in animal and human brains. We find that due to the significantly improved resolution and the large solid angle covered by the panel detectors, several issues that have been adequately solved for older generation scanners demand new attention for the HRRT, like acquiring and handling large amounts of data effectively, strategies for optimal reconstruction, shielding, and correction of random coincidences.

Neurotransmitter changes in dementia with Lewy bodies and Parkinson disease dementia in vivo

Objective: Although Parkinson disease with dementia (PDD) and dementia with Lewy bodies (DLB) show a wide clinical and neuropathologic overlap, they are differentiated according to the order and latency of cognitive and motor symptom appearance. Whether both are distinct disease entities is an ongoing controversy. Therefore, we directly compared patients with DLB and PDD with multitracer PET. Methods: PET with 18fluorodopa (FDOPA), N-11C-methyl-4-piperidyl acetate (MP4A), and 18fluorodeoxyglucose (FDG) was performed in 8 patients with PDD, 6 patients with DLB, and 9 patients with PD without dementia vs age-matched controls. Data were analyzed with voxel-based statistical parametric mapping and region of interest–based statistics. Results: We found a reduced FDOPA uptake in the striatum and in limbic and associative prefrontal areas in all patient groups. Patients with PDD and patients with DLB showed a severe MP4A and FDG binding reduction in the neocortex with increasing signal diminution from frontal to occipital regions. Significant differences between PDD and DLB were not found in any of the radioligands used. Patients with PD without dementia had a mild cholinergic deficit and no FDG reductions vs controls. Conclusions: Patients with dementia with Lewy bodies and Parkinson disease dementia share the same dopaminergic and cholinergic deficit profile in the brain and seem to represent 2 sides of the same coin in a continuum of Lewy body diseases. Cholinergic deficits seem to be crucial for the development of dementia in addition to motor symptoms. The spatial congruence of cholinergic deficits and energy hypometabolism argues for cortical deafferentation due to the degeneration of projection fibers from the basal forebrain.

Fast and robust registration of PET and MR images of human brain

In recent years, mutual information has proved to be an excellent criterion for registration of intra-individual images from different modalities. Multi-resolution coarse-to-fine optimization was proposed for speeding-up of the registration process. The aim of our work was to further improve registration speed without compromising robustness or accuracy. We present and evaluate two procedures for co-registration of positron emission tomography (PET) and magnetic resonance (MR) images of human brain that combine a multi-resolution approach with an automatic segmentation of input image volumes into areas of interest and background. We show that an acceleration factor of 10 can be achieved for clinical data and that a suitable preprocessing can improve robustness of registration. Emphasis was laid on creation of an automatic registration system that could be used routinely in a clinical environment. For this purpose, an easy-to-use graphical user interface has been developed. It allows physicians with no special knowledge of the registration algorithm to perform a fast and reliable alignment of images. Registration progress is presented on the fly on a fusion of images and enables visual checking during a registration.

Spreading depolarizations cycle around and enlarge focal ischaemic brain lesions

How does infarction in victims of stroke and other types of acute brain injury expand to its definitive size in subsequent days? Spontaneous depolarizations that repeatedly spread across the cerebral cortex, sometimes at remarkably regular intervals, occur in patients with all types of injury. Here, we show experimentally with in vivo real-time imaging that similar, spontaneous depolarizations cycle repeatedly around ischaemic lesions in the cerebral cortex, and enlarge the lesion in step with each cycle. This behaviour results in regular periodicity of depolarization when monitored at a single point in the lesion periphery. We present evidence from clinical monitoring to suggest that depolarizations may cycle in the ischaemic human brain, perhaps explaining progressive growth of infarction. Despite their apparent detrimental role in infarct growth, we argue that cycling of depolarizations around lesions might also initiate upregulation of the neurobiological responses involved in repair and remodelling.

Improved Herpes Simplex Virus Type 1 Amplicon Vectors for Proportional Coexpression of Positron Emission Tomography Marker and Therapeutic Genes

For the development of efficient and safe gene therapy protocols for clinical application it is desirable to determine the tissue dose of vector-mediated therapeutic gene expression noninvasively in vivo. The herpes simplex virus type 1 thymidine kinase gene (HSV-1-tk) has been shown to function as a marker gene for the direct noninvasive in vivo localization of thymidine kinase (TK) expression by positron emission tomography (PET). Using bicistronic or multicistronic gene-expressing cassettes with tk as the PET marker gene, the quantitative analysis of tk gene expression may indirectly indicate the distribution and the level of expression of linked and proportionally coexpressed genes. Here, we describe the construction and functional evaluation of HSV-1 amplicon vectors mediating proportional coexpression of HSV-1-tk as PET marker gene and the enhanced green fluorescent protein gene (gfp) as proof of principle and cell culture marker gene and the Escherichia coli cytosine deaminase (cd) as therapeutic gene. Several double-/triple-gene constructs expressing HSV-1-tk, gfp, and E. coli cd were engineered based on gene fusion or the use of an internal ribosome entry site (IRES). Functional analysis in cell culture (green fluorescent protein [GFP] fluorescence and sensitivity to the prodrugs ganciclovir [GCV] and 5-fluorocytosine [5-FC]) and Western blots were carried out after infection of proliferating rat 9L gliosarcoma and human Gli36 glioma cells with helper virus-free packaged HSV-1 amplicon vectors. To study the ability of PET to differentiate various levels of tk expression noninvasively in vivo, retrovirally transduced and selected populations of rat F98 and human Gli36dEGFR glioma cells with defined levels of proportionally coexpressed tk and gfp genes were grown as subcutaneous tumors in nude rats and nude mice, and tk imaging by PET was performed. To study HSV-1 amplicon vector-mediated gene coexpression in vivo, HSV-1 amplicon vectors bearing coexpression constructs were injected (4 x 10(7) to 1 x 10(8) transducing units) into subcutaneously growing Gli36dEGFR gliomas in nude animals, and tk imaging was performed 24 hr later. All vector constructs mediated GFP expression and sensitized 9L and Gli36 cells toward GCV- and 5-FC-mediated cell killing in a drug dose-dependent manner, respectively. The levels of gene expression varied depending on the location of the genes within the constructs indicating the influence of the IRES on the level of expression of the second gene. Moreover, functional proportional coexpression of the PET marker gene HSV-1-tk and the linked therapeutic E. coli cd gene was observed. In selected tumor cell populations, subtle IRES-dependent differences of tk gene expression could be noninvasively distinguished by PET with good correlation between quantitative assays for IRES-dependent attenuated GFP and TK expression in culture and in vivo. After infection of subcutaneously growing gliomas with HSV-1 amplicon vectors, various levels of TK expression were found ranging from 0.011-0.062 percentage injected dose per gram (%ID/g). These values were 4.0- to 5.7-fold lower than positive control tumor cells. TK expression could be imaged by PET in vivo even with the tk gene located at the weak position downstream from the IRES. In conclusion, these HSV-1 amplicon vectors carrying HSV-1-tk as PET marker gene and any linked therapeutic gene will serve an indirect noninvasive assessment of the distribution of therapeutic gene expression by PET. Monitoring the correlation between primary transduction and therapeutic efficiency of a given vector is highly desirable for the development of safe and efficient gene therapy and vector application protocols in clinical applications.

HeinzelCluster: accelerated reconstruction for FORE and OSEM3D

Using iterative three-dimensional (3D) reconstruction techniques for reconstruction of positron emission tomography (PET) is not feasible on most single-processor machines due to the excessive computing time needed, especially so for the large sinogram sizes of our high-resolution research tomograph (HRRT). In our first approach to speed up reconstruction time we transform the 3D scan into the format of a two-dimensional (2D) scan with sinograms that can be reconstructed independently using Fourier rebinning (FORE) and a fast 2D reconstruction method. On our dedicated reconstruction cluster (seven four-processor systems, Intel PIII@700 MHz, switched fast ethernet and Myrinet, Windows NT Server), we process these 2D sinograms in parallel. We have achieved a speedup >23 using 26 processors and also compared results for different communication methods (RPC, Syngo, Myrinet GM). The other approach is to parallelize OSEM3D (implementation of C Michel), which has produced the best results for HRRT data so far and is more suitable for an adequate treatment of the sinogram gaps that result from the detector geometry of the HRRT. We have implemented two levels of parallelization for our dedicated cluster (a shared memory fine-grain level on each node utilizing all four processors and a coarse-grain level allowing for 15 nodes) reducing the time for one core iteration from over 7 h to about 35 min.

Evaluation of Primary Brain Tumors Using 11C-Methionine PET with Reference to a Normal Methionine Uptake Map

11C-Methionine PET is a well-established technique for evaluating tumor extent for diagnosis and treatment planning in neurooncology. Image interpretation is typically performed using the ratio of uptake within the tumor to a reference region. The precise location of this reference region is important as local variations in methionine uptake may significantly alter the result, particularly for lesions at the border of gray and white matter. Selection of a reference region can be highly user dependant, and identifying a representative normal region may be complicated by midline or multifocal tumors. We hypothesized that current coregistration methods would enable interpretation of methionine PET images with reference to an averaged normal uptake map, allowing better standardization of scan analysis and increasing the sensitivity to tumor infiltration, particularly of white matter regions. Methods: A normal methionine uptake map was prepared from the normal hemispheres of 20 scans performed on patients with benign or low-grade lesions. Affine and nonlinear coregistration algorithms were evaluated for spatial normalization of the images to a previously developed PET template. A standardized method for applying the normal uptake map in brain tumors was developed and evaluated in a sample of 18 scans (6 grade II, 6 grade III, and 6 grade IV gliomas). Tumor extent was compared with that derived from a mirrored contralateral reference region method. Correlation coefficients were calculated between the uptake ratios for tumor to normal uptake map versus tumor to mirrored reference region. Results: “RatioMap” images depicting voxel-by-voxel ratios of a patient scan to the normal uptake map revealed increased methionine uptake in white matter regions that could not be identified using the standard method. Uptake ratios within the tumor varied slightly with the normalization methods used but correlated closely with the ratio to a single reference value. Nonlinear coregistration with median ratio intensity normalization gave the strongest correlation (r = 0.97, P < 0.001, n = 17). Conclusion: Evaluation of methionine PET data with reference to normal uptake data may improve sensitivity to white matter infiltration. The tumor uptake ratios obtained correlated closely with a standard reference value technique, whereas the described method allowed for better standardization of the image analysis.

Distinct spatiotemporal patterns of spreading depolarizations during early infarct evolution: evidence from real-time imaging

Experimental and clinical studies indicate that waves of cortical spreading depolarization (CSD) appearing in the ischemic penumbra contribute to secondary lesion growth. We used an embolic stroke model that enabled us to investigate inverse coupling of blood flow by laser speckle imaging (CBFLSF) to CSD as a contributing factor to lesion growth already in the early phase after arterial occlusion. Embolization by macrospheres injected into the left carotid artery of anesthetized rats reduced CBFLSF in the territories of the middle cerebral artery (MCA) (8/14 animals), the posterior cerebral artery (PCA) (2/14) or in less clearly defined regions (4/14). Analysis of MCA occlusions (MCAOs) revealed a first CSD wave starting off during ischemic decline at the emerging core region, propagating concentrically over large portions of left cortex. Subsequent recurrent waves of CSD did not propagate concentrically but preferentially circled around the ischemic core. In the vicinity of the core region, CSDs were coupled to waves of predominantly vasoconstrictive CBFLSF responses, resulting in further decline of CBF in the entire inner penumbra and in expansion of the ischemic core. We conclude that CSDs and corresponding CBF responses follow a defined spatiotemporal order, and contribute to early evolution of ischemic territories.

In-vivo visualization of tumor microvessel density and response to anti-angiogenic treatment by high resolution MRI in mice

Purpose Inhibition of angiogenesis has shown clinical success in patients with cancer. Thus, imaging approaches that allow for the identification of angiogenic tumors and the detection of response to anti-angiogenic treatment are of high clinical relevance. Experimental Design We established an in vivo magnetic resonance imaging (MRI) approach that allows us to simultaneously image tumor microvessel density and tumor vessel size in a NSCLC model in mice. Results Using microvessel density imaging we demonstrated an increase in microvessel density within 8 days after tumor implantation, while tumor vessel size decreased indicating a switch from macro- to microvessels during tumor growth. Moreover, we could monitor in vivo inhibition of angiogenesis induced by the angiogenesis inhibitor PTK787, resulting in a decrease of microvessel density and a slight increase in tumor vessel size. Conclusions We present an in vivo imaging approach that allows us to monitor both tumor microvessel density and tumor vessel size in the tumor. Moreover, this approach enables us to assess, early-on, treatment effects on tumor microvessel density as well as on tumor vessel size. Thus, this imaging-based strategy of validating anti-angiogenic treatment effects has high potential in applications to preclinical and clinical trials.

Integration of Functional Data (PET) into Brain Surgery Planning and Neuronavigation

There are distinct advantages of including functional data from PET into surgery planning and neuronavigation: tumor delineation and localization of functional areas. Preoperative intervention planning in brain tumours poses many additional challenges if the goal is to include functional data from PET for neuronavigation: co-registration of multimodal image data (MRI and PET) is conceptually the most demanding task. However, we find that associated problems of visualization, data transfer, documentation and, in general terms, quality assurance are often underestimated. We address some of these problems, including visual inspection of co-registration accuracy, preparation of image data, DICOM capabilities and “electronic documentation” [1].