15th European Molecular Imaging Meeting
supported by:

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

Late-breaking Pitches | Imaging Strategies

Session chair: Franz Schilling (Munich, Germany); Fabian Kiessling (Aachen, Germany)
Shortcut: PS 15
Date: Thursday, 27 August, 2020, 10:00 a.m. - 11:30 a.m.
Session type: Parallel Session


Abstract/Video opens by clicking at the talk title.

10:00 a.m. PS 15-01

Increased brain metabolism after Western style diet observed by in vivo PET imaging

Marilena Poxleitner1, Sabrina H. L. Hoffmann1, Philipp Knopf1, Laura Kuebler1, Sabrina Buss1, Simon Freisinger1, Vera Joerke1, Andreas Maurer1, Gerald Reischl1, Kristina Herfert1, Nicolas Bézière1, Bernd J. Pichler1

1 University of Tuebingen, Preclinical Imaging and Radiopharmacy, Tuebingen, Germany


Diet-induced obesity is not only a major risk factor for chronic diseases, but is discussed to affect central nervous system (CNS) functions as well. Therefore, diet-induced metabolic chronic inflammation in peripheral organs, which lead to the release of inflammatory mediators could potentially affect the CNS and e.g. increase the susceptibility for neurodegenerative disorders like Alzheimer Disease (AD). This project aims to determine the influence of a western style diet on brain metabolism in healthy and AD mice using a multi-PET tracer approach.


2 months old APPPS1 (PS1) mice and C57BL/6J (WT) were fed with either a western diet (WD) or a normal diet (ND) for 6 months. At 8 months of age, 60 min dynamic PET scans using [18F]FDG, a marker for glucose metabolism, [18F]FTHA, a marker for fatty acid metabolism and [18F]GE-180, which binds to the translocator protein (TSPO) marking activated microglia were performed on consecutive days in the same animals. Whole brain time activity curves were generated and mean SUV values were calculated for the last 30-60 min post tracer injection. Flow cytometry analysis (FC) of brain tissue and a cytokine profiling (CP) of plasma samples were performed 7 days after the final PET scan. Baseline FC and cytokine experiments were performed with 2 months old PS1 (n=5) and WT (n=7) controls.


Under ND, SUVs of WT and PS1 mice were comparable. WD increased [18F]FDG uptake in PS1 mice (SUVND=1.60±0.36 n=7, SUVWD=2.05±0.38 n=8, p=0.005), but not in WT mice (SUVND= 1.39±0.25 n=11, SUVWD=1.60±0.22 n=7), pointing to a dysregulated glucose metabolism in PS1 mice. [18F]FTHA uptake was affected in both, WT (SUVND=0.44±0.05 n=9, SUVWD=0.58±0.07 n=8, p<0.001) and PS1 mice (SUVND=0.45±0.04 n=8, SUVWD=0.56±0.09 n=7, p=0.01) revealing genotype-independent WD-induced fatty acid metabolism alterations. [18F]GE-180 uptake was higher in PS1 under ND (SUVPS1=0.82±0.08 n=7, SUVWT=0.58±0.09 n=9, p=0.001). This effect was unchanged with the WD (SUVPS1=0.86±0.14 n=10, SUVWT=0.70±0.12 n=9, p=0.02). Thus, the WD does not alter microglial activation. No changes between baseline PS1 and WT brains with FC and CP were detected, while WD treatment resulted in a reduction in CD4+ cells in WT and PS1 mice. Inflammatory plasma cytokine levels in WD mice were elevated, revealing systemic metabolic changes.


Here, we could show that a 6-month WD affects brain metabolism in mice and that this effect was not related to microglia activation. However, we observed immune cell changes in WD brains, suggesting that circulating T cells may play a role in metabolically induced brain inflammation due to permanent overnutrition. Ex vivo histology is ongoing to validate the impact of CNS immune cell contribution.

Keywords: Alzheimer, Western Diet, PET
10:10 a.m. PS 15-02

Graph theory analysis of brain networks in absence epilepsy

Lydia Wachsmuth1, Katharina Kemper1, Franziska Albers1, Henriette Lambers1, Maia Datunashvili2, Annika Lüttjohann2, Silke Kreitz3, Thomas Budde2, Cornelius Faber1

1 University of Münster, Clinical Radiology, Münster, Germany
2 University of Münster, Physiology I, Münster, Germany
3 University of Erlangen, Experimental and Clinical Pharmacology and Toxicology, Erlangen, Germany


Genetic Absence Epilepsy Rats from Strasbourg (GAERS) represent an established animal model for childhood absence epilepsy1. Seizures show characteristic pattern of generalized brain activity, spike and wave discharges (SWD), in electrophysiological or optical recordings. We performed graph theory analysis of resting-state (rs)fMRI data acquired from GAERS, before and during SWDs, and from nonepileptic control animals (NEC) under Narcolept anesthesia. With subsequent electrophysiological recordings we evaluated whether connectivity changes were associated with alterations on cellular level.


30 min BOLD rsfMRI of adult GAERS (n=9) before and during SWDs and NEC (n=8) were performed with GE-EPI at 9.4 T (Bruker Biospec) under Narcolept anesthesia. Brain state was identified by simultaneous Ca2+ recordings (OGB-1) in motor cortex. Data was preprocessed (SPM) and registered to an anatomical template (MagnAn2). Cross-correlation matrices of regional time courses were averaged and thresholded (approx. 10 connections per brain region). Global and local network parameters were calculated and topology displayed with Gephi3. Statistically significant different connections between brain regions were determined with NBS statistics4. We also performed slice recordings in voltage and current clamp mode and pyramidal- and interneuron specific immunostaining in retrosplenial cortex (RS).


Normalized clustering coefficient, path length and small worldness are similar on group level. Functional groups gather within distinct communities. Comparison between brain states in GAERS (fig. 1) reflects acute changes during seizures: Cortical community separates from other communities. During seizures, more statistically significant connections are found between limbic and thalamic structures. High betweenness centrality and degree indicate that RS gains hub status. Comparison between NEC and preSWD GAERS (fig. 2) indicates longterm changes: Five of top10 central betweenness structures in NEC belong to the sensorimotor cortex. Thalamic and basal ganglia structures are better connected in NEC. Hypothalamus (Ht) has hub status in preSWD GAERS. Slice experiments in RS reveal higher excitability of layer V pyramidal neurons, higher frequency of inhibitory postsynaptic currents and increased immunoreactivity for parvalbumin interneurons of GAERS.


Global small world topology is preserved in epileptic animals. Local parameters change in a complex fashion. In particular, altered connectivity of the retrosplenial cortex indicates an important role of this region within the thalamocortical network during seizures. Interestingly, subsequent electrophysiological slice experiments reveal alterations in synaptic function and intrinsic neuronal firing pattern in this brain region.

[1] Depaulis A, David O, Charpier S, 2016, The genetic absence epilepsy rat from Strasbourg as a model to decipher the neuronal and network mechanisms of generalized idiopathic epilepsies, J Neurosci Methods 260:159-74
[2] Biocom, Uttenreuth
[3] https://gephi.org/
[4] Zalesky A, Fornito A, and Bullmore ET, 2010, Network-based statistics: identifying differences in brain networks, Neuroimage, 53, 1197-1207
Network topology in GAERS, before and during seizures
Comparison between GAERS a) in preSWD and b) SWD state. Nodes are arranged according to a force-based algorithm: Strongly connected nodes appear close together. Node color and ellipsis indicate community affiliation. One cortical community (orange), a community (SWD only) with cortical and basal ganglia structures (yellow), two mostly thalamic communities (blue, purple), and one community with basal ganglia and limbic structures (green). Size of nodes codes betweenness centrality. Statistically significant (a = 0.05) stronger connections, c) preSWD and d) SWD, are overlaid on glass brains.
Network topology in NEC and GAERS
Comparison between a) NEC and b) preSWD GAERS. Nodes are arranged according to a force-based algorithm: Strongly connected nodes appear close together. Node color and ellipsis indicate community affiliation: One cortical community (orange), two mostly thalamic communities (blue, purple), and two communities with basal ganglia and limbic structures (green). Size of nodes codes betweenness centrality. Statistically significant stronger connections, c) NEC and d) preSWD, are overlaid on glass brains.
Keywords: resting state fMRI, epilepsy, rat
10:20 a.m. PS 15-03

Season dependent auditory processing of innate vocalizations in a songbird model for seasonal adult neuroplasticity.

Nicholas Vidas-Guscic1, Elisabeth Jonckers1, Jasmien Orije1, Julie Hamaide1, Gaurav Majumdar1, Annemie Van der Linden1

1 University of Antwerp, Bio-Imaging Lab, Wilrijk, Belgium


Innate vocalizations from the offspring (eg. baby cry) are perceived differently depending on the hormonal status of the receiver1.  A model to study this is the European Starling, a seasonal songbird which goes through a photoperiod induced seasonal hormonal cycle. We investigated whether nestling begging calls, a typical breeding season signal, is perceived and processed in a different way in and outside the breeding season. By adjusting auditory processing to the behavioral relevance of the incoming signals, communication in a continuously changing environment is facilitated2.


We applied auditory functional MRI on anesthetized European starlings (11 male;8 female) in artificial induced breeding (spring) and non-breeding (fall) physiologies on a 7T PharmaScan MR system using a T2-weighted turbo-RARE sequence (Effective echo-time= 50.60ms, Repetition time= 2s). Fifteen sagittal slices (1.0mm) were recorded (resolution 0.33x0.67mm2). Birds were stimulated with starling nestling begging calls, synthetic pure tones and Starling warble motifs, chosen as control stimulus with a non-differing seasonal processing3 . Each stimulus was presented 42 times; interleaved by resting blocks. Standard preprocessing and statistical voxel-based analyses using a repeated measures ANOVA for each stimulus with season as within factor and sex as between factor were performed in SPM12.


Simple t-contrasts (stimulus>rest) demonstrated activation in the bilateral higher auditory nuclei Field L (primary auditory cortex) and NCM/CMM (secondary auditory cortex) for all stimuli in both seasons (Puncorrected<0.001). Overall comparison of the stimuli showed significantly higher BOLD activation for Begging calls compared to the other stimuli. The contrast Breedingbegging > Non-breedingbegging resulted in a significant difference in caudal-ventral NCM (PFWE=0.006, Tpeak=5.54) (Figure) showing differential processing depending on the relevance of the stimulus. The control stimuli with invariable relevance did not show a seasonally changing response. No significant effects of sex sex × season interaction could be found in any auditory region. This matches with the fact that both sexes are engaged in parental care (incubating the eggs as well as feeding of off-spring until fledgling).


Our findings imply that photoperiod induced seasonal physiologies generate differential neural responses in the receiver. Such differential processing leads to an improved perception of communication via a mechanism that appears to be equally important for both sexes. The regions elucidated with fMRI can be further investigated with molecular and histological techniques to unravel the underlying cellular mechanism.


EJ  and GM were funded with a Post Doc grant of the FWO Flanders. JO was a FWO Flanders PhD fellow. 


[1] Swain, JE 2008, Baby stimuli and the parent brain: functional neuroimaging of the neural substrates of parent-infant attachment., Psychiatry, 5(8), 28-36.
[2] Sridharan, D, et al. 2014, Selective attention in birds., Curr Biol., 24(11), 510-3. 
[3] De Groof, G, et al. 2013, Functional changes between seasons in the male songbird auditory forebrain., Front Beh Neurosci, 7, 196.
[4] De Groof, G, et al. ,2016, A three-dimensional digital atlas of the starling brain. Brain Struct. Funct., 221(4),1899-909.
Neural activation for begging call changes significantly between seasons.
Left: Statistical map displaying voxels with significantly higher activation in the breeding season (three different orientations). Significant voxels have been superimposed on the starling MRI atlas4  rmANOVA: N= 11 males & 8 females, Puncorrected<0.001, kcluster>5). Right: Crosshairs indicate the voxel with the highest T-value. Color scaleof T-values is presented on the right. Right: 3D color-coded representation of starling auditory cortex.
Keywords: functional MRI, Auditory Processing, hormones
10:30 a.m. PS 15-04

Croconaine-derived CR780 nanoparticles enable efficient optoacoustic imaging of deep brain tumor

Nian Liu1, 2, Vipul Gujrati1, 2, Jaber Malekzadeh-Najafabadi1, 2, Juan Pablo Fuenzalida Werner1, Yuanhui Huang1, 2, Longguang Tang3, Jaya Prakash1, Zhenyue Chen1, Uwe Klemm1, Andre C. Stiel1, Gabriele Mettenleiter4, Michaela Aicheler4, Axel Walch4, Daniel Razansky1, Michael Sattler5, Vasilis Ntziachristos1

1 Helmholtz Zentrum München, Institute of Biological and Medical Imaging, Munich, Germany
2 Technische Universität München, Chair of Biological Imaging, TranslaTUM, Munich, Germany
3 Xiamen University, State Key Laboratory of Molecular Vaccinology and Molecular Diagnostics & Center for Molecular Imaging and Translational Medicine, School of Public Health, Xiamen, Germany
4 Helmholtz Zentrum München, Research Unit Analytical Pathology, Munich, Germany
5 Technische Universität München, Center for Integrated Protein Science Munich at Department Chemie, Munich, Germany


Optoacoustic (OA) imaging can provide functional images of the brain by resolving hemoglobin contrast.1 However, visualization of brain tumors could benefit from the development of agents that exhibit high OA contrast, high photostability, tumor targeting ability, optimal clearance from the circulation and low toxic side effects.2,3 We introduce croconaine dye based ultra-small nanoparticles (CR780RGD-NPs) as a highly efficient contrast agent for targeted deep brain tumor optoacoustic imaging.


We synthesised croconaine backbone-derived CR780 dye conjugated with polyethylene glycol and cancer targeting RGD peptide, that self-assemble in aqueous media to form ultra-small nanoparticles (CR780RGD-NPs). We did the physical-chemical characterization for CR780RGD-NPs (TEM, absorption spectrum, optoacoustic conversion efficiency (OGE), photostability) and tested the OA imaging depth using tissue-mimicking phantoms. Next, we carried out multispectral optoacoustic tomography (MSOT) imaging using orthotopic glioblastoma models after i.v. injection of CR780RGD-NPs,. Finally, we did Hematoxylin-eosin (H&E) staining of brain slices.


CR780RGD-NPs exhibit 25 nm average size and strong near-infrared absorption (Fig. 1a, 1b). They have high OGE and high photostability, better than FDA approved ICG (Fig. 1c).  CR780RGD-NPs with very low dose concentration could show strong OA signal from 5 mm depth in brain tissue, indicate that CR780RGD-NPs are suitable for deep tissue imaging. Due to ultra-small particle size (average size 25 nm) and active targeting ability, CR780RGD-NPs could effectively reach to the brain tumor by enhanced permeability and actively accumulate in the tumor region to show strong and persistent OA contrast (Fig. 1d). H&E staining of brain slices confirmed the existence of the brain tumor.


In summary, our results show the powerful capabilities of MSOT and the developed RGD peptide modified CR780-NPs for targeted OA imaging of brain tumor. The ultra-small size of NPs ensures efficient disrupted BBB penetration, while tumor-targeting ability, high OGE and photostability ensure enhanced OA contrast from the nano-agent.


This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 694968 (PREMSOT). The research leading to these results was supported by the Deutsche Forschungsgemeinschaft (DFG), Germany (Gottfried Wilhelm Leibniz Prize 2013, NT 3/10–1) as well as by the DFG as part of the CRC 1123 (Z1).

[1] Ntziachristos, V, Razansky, D, 2010, ‘Molecular Imaging by Means of Multispectral Optoacoustic Tomography (MSOT)’, Chem. Rev, 110, 2783-2794.
[2] Furtado, D, Björnmalm, M, Ayton, Scott, Bush, A, Kempe, K, Caruso, F, 2018, ‘Overcoming the Blood-Brain Barrier: The Role of Nanomaterials in Treating Neurological Diseases’, Adv. Mater, 30, 1801362.
[3] Gao, D, Hu, D, Liu, X, Sheng, Z, Zheng, H. 2019, ‘Recent advances in functional nanomaterials for photoacoustic imaging of glioma’, Nanoscale Horizons, 4, 1037-1045.
Figure 1

(a) Schematic of the CR780RGD-NPs synthesis for OA imaging of brain tumour. (b) Absorption spectrum of CR780RGD-NPs. (c) OA signal intensity of ICG, CR780, CR780-NPs and CR780RGD-NPs at different concentration. (d) In vivo multi-spectral optoacoustic tomography (MSOT) imaging of CR780RGD-NPs in U87 bearing orthotopic glioblastoma models. Signals are shown only for the brain region. Scale bar, 5 mm. 

Keywords: Brain tumor, Croconaine nanoparticles, Optoacoustic, MSOT, high penetration depth
10:40 a.m. PS 15-05

Radiofluorinated gases as regional ventilation markers: Application to a rat model of acute lung inflammation

Unai Cossío1, Vanessa Gómez-Vallejo1, Pedro Ramos-Cabrer1, 2, Jordi Llop1, 3

1 CIC biomaGUNE, Basque Research and Technology Alliance (BRTA), San Sebastian, Spain
2 Ikerbasque, Basque Foundation for Science, Bilbao, Spain
3 Centro de Investigación Biomédica en Red, Enfermedades Respiratorias - CIBERES, Madrid, Spain


Imaging methods visualizing local areas of impaired ventilation may become a powerful tool in the early/differential diagnose of lung diseases. Currently, clinical ventilation studies are performed with SPECT using radiolabelled aerosols. However, aerosols show deposition in major airways and long residence time in lungs. Additionally, SPECT has intrinsic limitations in terms of sensitivity and spatiotemporal resolution. Here, we investigate the use of [18F]CF4 (gas) as a PET ventilation marker in an animal model of impaired lung ventilation.


[18F]CF4 was produced as recently reported [1]. Ventilation studies were performed both in healthy rats and in a rat model of lung inflammation, induced by intratracheal administration of lipopolysaccharide (LPS). Dynamic 10-min PET images under continuous inhalation of [18F]CF4 were obtained in list mode at t=4 and t=48 hours after LPS administration. [18F]FDG-PET images were also obtained after the finalisation of ventilation studies. Images were reconstructed by 3DOSEM iterative algorithm. For ventilation studies, voxel-by-voxel analysis was carried out to determine the concentration of the radioactivity in the different regions of the lungs, normalised to the concentration in the trachea (reference region). Standard uptake values (SUV) in lungs were determined for [18F]FDG-PET images.


Ventilation studies in healthy animals showed uniform distribution of the radiofluorinated gas (Figure 1a) and fast elimination of the radioactivity after discontinuation of the administration. For LPS-treated rats, no ventilation defects were observed at 4 hours after LPS administration. However, at 48 hours, hipoventilated regions could be clearly visualised (Figure 1c). The mean concentration of [18F]CF4 in the lungs (relative to the concentration in the trachea) in LPS-administered animals was significantly lower than that obtained for the control group, and voxel-by-voxel analysis showed a clear displacement of the histogram to lower values (Figure 1b). CT images confirmed that hipoventilated areas correspond to edema formation (Figures 1d-e). [18F]FDG images showed 2-fold increase in SUV values in the lungs of LPS-treated rats at 4h (Figure 2), and high localised uptake in the regions affected by edema at t=48 h (Figure 1f).


[18F]CF4 is an appropriate marker of regional lung ventilation and may find application in the early diagnose of acute lung disease.


The work was funded by the Spanish Ministry of Economy and Competitiveness, grant number CTQ2017-87637-R.

[1] Gómez-Vallejo V, Lekuona A, Baz Z, Szczupak B, Cossío U, Llop J. "Ion beam induced 18F-radiofluorination: straightforward synthesis of gaseous radiotracers for the assessment of regional lung ventilation using positron emission tomography". Chemical Communications. 2016, 52(80), 11931.
Figure 1
a, c) Maximal Intensity Projection (MIP) PET-CT images of a control rat (a) and a LPS-treated rat (c) 48 h after treatment (scale bar: 0-50 kBq/cc); b) histograms corresponding to control (blue line) and LPS-treated (red line) animals. Values represent concentration of radioactivity in the voxel, relative to average concentration of radioactivity in the trachea; d, e) coronal and axial CT slices, respectively, where regions with edema are visible; f) axial PET-CT image of a LPS-treated rat 48h after treatment. High localised uptake of [18F]FDG is visible in the regions affected by edema.
Figure 2
Maximal Intensity Projection (MIP) [18F]FDG-PET images (coronal views) overlaid with CT images (representative slices) of a control rat (a) and a LPS-treated rat (b) 4 hours after treatment.
Keywords: Lung ventilation, PET-CT, LPS, [18F]CF4, voxel-by-voxel analysis
10:50 a.m. PS 15-06

Efficient integration of a SPECT-CT small animal scanner based on photon-counting technology

Martin P. Pichotka1, 2, Moritz Weigt4, Peter Manek2, 7, Peter Rubovic2, Benedikt Bergmann2, Peter Burian5, 2, Dominik von Elverfeldt3, Michael Mix1, 6

1 University Medical Center — University of Freiburg, Nuclear Medicine, Freiburg, Germany
2 CTU Prague, Institute of Experimental and Applied Physics, Praha, Czech Republic
3 University Medical Center — University of Freiburg, Department of Radiology - Medical Physics, Freiburg, Germany
4 speCTive GmbH, R&D, Freiburg, Germany
5 University of West Bohemia, Department of electronics, Plzen, Czech Republic
6 Stellenbosch University, Div. of nuclear medicine, Cape Town, South Africa
7 University College London, Dep. of high energy physics, London, United Kingdom


With the fast evolution of Hybrid-Pixel-Detector technology new imaging applications are coming within reach. The current contribution aims to demonstrate the feasibility of a multi-modal SPECT/CT system (fig. 1a), featuring material discrimination as well as Compton-camera based collimator-free emission imaging. The proposed system uses the same detector array for emission- and transmission-CT, making the design economical while facilitating image registration. Alongside the system design, efficient routines for data handling and volume reconstruction for both modalities have been developed.


The current study utilizes detectors based on the Medipix/Timepix ASIC family. In particular the Timepix3 ASIC permits for all operational modes necessary for combined SPECT-CT imaging. In order to handle the high photon flux in radiographic applications the detector is operated in energy-thresholded counting mode, allowing for energy binned CT acquisition. The photon starvation within CT energy channels with increasing bin density is counteracted by employing very robust iterative volume reconstruction routines. Compton imaging is performed in data-driven single event mode, featuring 10bit energy resolution. The 1.5ns timing resolution in this mode facilitates coincidence assignment but also allows for 3D event tracking within a single semiconductor sensor by drift time analysis.


The spectroscopic performance in transmission CT is demonstrated in fig. 1b-c, where a tissue equivalent micro-CT phantom is decomposed to RGB channels. Here different materials can be clearly distinguished by color, representing the respective absorption spectra. SPECT imaging, using the same detector technology, is shown in fig. 2. Fig. 2a shows a combined SW tool, which simultaneously allows for device simulation and optimization(shown in fig. 2b for primary photon energy and device energy resolution), as well as volume reconstruction from experimental data. As observed in fig. 2c, with the resulting Compton-camera 3D information can be extracted even from a single view angle. While the measurements resulting in figs. 1b and 2c have been obtained independently, the fact that both have been done with the same detector technology clearly shows that both modalities can be integrated into a common system, making dual use of a single detector system.


With the recent advent of multi-modal scanners in medical applications the benefits of correlating 3D data from functional and anatomical imaging evermore have become apparent. The higher cost sensitivity as well as space limitation in animal scanners render the prospect of integrated SPECT/CTs, using a single detector array and gantry, very attractive. The current contribution demonstrates the feasibility of such a system by available technology

AcknowledgmentThis contribution has been supported by Eurostars project P3D, ProjID: 10998
Multi-modal SPECT/CT system - design concept and spectroscopic CT performance

Left: Conceptual design of the single gantry SPECT/CT scanner. The system optionally can be fitted with a single- or multi-layer HPD detector array.

Center: Single channel reconstruction of a Gammex tissue equivalent CT phantom.

Right: RBG representation of the same phantom in energy binned reconstruction. The bin edges are placed at 12/18/23/30keV respectively. Sensor: 1mm CdTe, W-anode transmission tube, 90kVp, .1mA, 200 projections, .2s each.

HPD based Compton camera - development of software-tools and experimental verification

Top left: Unified integration of software tools for volume reconstruction and system simulation.

Bottom left: Simulation study on the angular resolution of the Compton imager in dependence of the detectors energy resolution and the primary photon energy.

Right: Source reconstruction from a single exposure, featuring 50k coicindent events from a  57Co capillary source.

Keywords: Compton camera, spectroscopic CT, multi-modal imaging, SPECT/CT, hybrid pixel detector
11:00 a.m. PS 15-07

Integrated dispersion correction of simultaneously recorded PET and dynamic contrast-enhanced MRI arterial input functions using an extracorporeal circulation in mice

Philipp Backhaus1, 2, 3, Klaus P. Schäfers1, Sven Hermann1, Lydia Wachsmuth3, Cornelius Faber3, Michael Schäfers1, 2, Florian Büther1, 2

1 University of Münster, European Institute for Molecular Imaging (EIMI), Münster, Germany
2 University Hospital Münster, Department of Nuclear Medicine, Münster, Germany
3 University of Münster, Translational Research Imaging Center (TRIC), Münster, Germany


Compartmental modeling in small animal PET and MRI can serve to better judge the performance of radiotracer candidates and the translational potential of contrast agent-driven perfusion studies. Precise recording of the arterial input function (AIF) is a key prerequisite for modeling, but its acquisition is very challenging in small animals. So far, no method for simultaneous recordings of PET radiotracers and MR contrast agent AIFs has been reported in mice. This impedes exploring the potential synergistic effects for dual AIF recordings and integrated modeling in PET/MRI imaging in mice.


We recently introduced a novel experimental setup allowing to record DCE-MRI AIFs in mice using an extracorporeal circulation. The shunted blood perfuses two reservoirs that reside in the field of view of a 9.4 T Bruker MRI for dynamic recording of the arterial contrast agent concentration. A downstream MR-compatible chamber with a plastic scintillator (b-microprobe) was used to record the arterial radiotracer concentrations. Either [18F]-FDG or [18F]-PSMA1007 was mixed with Gadobutrol (Gadovist®) and injected intravenously. Dispersion correction was validated in separate experiments by inserting a b-microprobe into the aortic arch. The impact of dispersion correction on modeling cerebral glucose consumption was assessed by recording mice in a Mediso 1T PET/MRI after [18F]-FDG injection.


The peak concentrations of simultaneously injected radiotracers and MR contrast agents showed very good agreement demonstrating quantitative robustness of our setup for both modalities. Dispersion at the different measurement sides (aortic arch, two MR reservoirs and measuring chamber) could be described by a common model allowing to perform individual based integrated multi-modal dispersion correction. Pharmacokinetic modeling of cerebral glucose consumption demonstrated the impact of dispersion uncertainties on modeling parameters.


We report the first simultaneous recordings of PET radiotracer and MRI contrast agent AIFs in mice by using a novel extracorporeal circulation model. We could demonstrate potential synergistic effects for integrated multi-modal dispersion correction and set the methodological basis for integrated PET/MRI AIF recordings and pharmacokinetic modeling.

Keywords: AIF, PET/MRI, Dispersion
11:10 a.m. PS 15-08

Lipid imaging for diagnosis of neuroinflammation after preclinical stroke

Ana Joya1, 2, Ibai Calvo3, Daniel Padro4, Juan José Gutierrez1, 3, Sandra Plaza-García4, Vanessa Gómez-Vallejo2, Pedro Ramos-Cabrer4, 5, Lucía Martín3, Krishna R. Pulagam2, Unai Cossío2, Jordi Llop2, José Andrés Fernández3, Abraham Martín1, 5

1 Achucarro Basque Center for Neuroscience, Leioa, Spain
2 CIC biomaGUNE, Radiochemistry and Nuclear Imaging, Molecular Imaging Unit, San Sebastian, Spain
3 University of the Basque Country, Dep. of Physical Chemistry, Leioa, Spain
4 CIC biomaGUNE, Magnetic Resonance Imaging, Molecular Imaging Unit, San Sebastian, Spain
5 Ikerbasque Basque Foundation for Science, Bilbao, Germany


The formation of lipid bodies (LBs) stands in agreement with progressive changes underwent by microglia and infiltrated macrophages after inflammation1. Despite this, the in vivo imaging of lipids as surrogate markers of neuroinflammation has been scarcely evaluated so far. For this reason, this study pursues to explore the multidisciplinary development of a novel and alternative method for the diagnosis of neuroinflammation by in vivo proton nuclear magnetic resonance spectroscopy (1H-MRS) and ex vivo Matrix-Assisted Laser Desorption/Ionization (MALDI) imaging in preclinical stroke.


In vivo imaging evaluation of lipids was carried out using PET imaging with the radiotracer [18F]DPA-714 in combination with MRI imaging before and at days 1, 3, 7, 14, 21 and 28 after transient middle cerebral artery occlusion (MCAO) in rats (n=10). PET imaging was conducted to measure the expression of TSPO as gold-standard imaging method to study neuroinflammation. Likewise, MRI studies were used to determine the volume of infarction and detection of lipids in the lesion using T2W-MRI and 1H-MRS, respectively. Finally, the ex vivo evaluation of lipid content was evaluated using MALDI-imaging, immunohistochemistry and confocal microscopy at day 7 after cerebral ischemia (n=4).


PET imaging with [18F]DPA showed a progressive uptake during the day 3 to 7 followed by a progressive decline from day 15 to day 28 after ischemia onset. In the other hand, 1H-MRS imaging was able to detect an increase of some mobile fatty acyl chains of tissue triacyilglycerides (0.9, 1.2, 2.0 and 5.3ppm) forming lipid bodies (LBs) in the region of the ischemic lesion at days 3 and 7 after ischemia (Figure 1). Additionally, MALDI-imaging evaluation showed the overexpression of some lipid families such as glucosylceramide, phosphatidylglycerol, phosphatidylinositol, and sphingomyelin in the ischemic brain at day 7 after stroke (Figure 2). Finally, the formation of LBs after ischemia was detected in activated microglia and infiltrated macrophages using immunoflurorescence staining and confocal microscopy with specific markers for lipids (Bodipy) and microglia/macrophages (CD11b and Iba1).


These results showed that 1H-MRS and MALDI-imaging provide accurate quantitative information on the time course of lipid formation as a marker for microglial/macrophage activation in experimental stroke. Hence, the exploration of this neuroinflammatory imaging methodology is a revolutionary diagnostic tool for the in vivo detection and treatment of neuroinflammation with a high degree of applicability in the routine neurological care.

AcknowledgmentThe authors would like to thank to the Spanish Ministry of Science and Innovation (RYC-2017-22412 and SAF2014-54070-JIN) for financial support.
[1] Gasparovic, C.;  Rosenberg, G. A.;  Wallace, J. A.;  Estrada, E. Y.;  Roberts, K.;  Pastuszyn, A.;  Ahmed, W.; Graham, G. D., Magnetic resonance lipid signals in rat brain after experimental stroke correlate with neutral lipid accumulation. Neuroscience letters 2001, 301 (2), 87-90
Figure 1.
Detection of lipids after stroke using 1H-MRS
Figure 2

MALDI-imaging of lipids at day 7 after cerebral ischemia.

Keywords: lipids, cerebral ischemia, MRS, MALDI, PET
11:20 a.m. PS 15-09

Multimodal anatomical and functional MRI combined with feature selecting approaches: an ideal analysis framework for extracting characteristics of responders vs. non-responders showcased in Crohn’s disease patients

Mageshwar Selvakumar1, Marina Sergeeva1, Laura C. Konerth1, Julie Roesch2, Arnd Dörfler2, Raja Athreya3, Markus Neurath4, Andreas Hess1

1 Friedrich Alexander Universität Erlangen Nürnberg, Institut für Pharmakologie und Toxikologie, Erlangen, Germany
2 Universitätsklinikum Erlangen, Neuroradiologie, Erlangen, Germany
3 Universitätsklinikum Erlangen, Medizin I, Erlangen, Germany
4 Universitätsklinikum Erlangen, Lehrstuhl für Innere Medizin I, Erlangen, Germany


Crohn’s disease (CD), a subtype of inflammatory bowel disease, is caused by immune-mediated inflammation in the gastrointestinal tract. However, increasing evidence suggests an important cross-talk between gut and brain[1]. In-depth studies of brain functions require multimodal imaging procedures that cover different anatomical but also functional modalities. This results in high-dimensional data that require adaptive statistical analysis strategies. Programs that do this in a flexible but user-friendly manner are not available. Therefore, we here present an appropriate analysis framework.


17 CD patients underwent 3 MRI sessions on a SIEMENS 3T MRI: before (M1), 24 h (M2), and 27 days (M3) after treatment. 33 matching controls were scanned identically. Each session consisted of several scans: T1 MPRAGE and standard DTI scan, BOLD EPI functional resting-state as well as repetitive painful abdominal compression stimulation. All data underwent standard analysis pipelines: anatomy: freesurfer[2] and VBM, functional data: MRSA[3] and stimulation BOLD[2]. For multimodal analysis we implemented a flexible framework in R, MARIA (MAgnetic Resonance Imaging data Analysis) with several packages e.g. Rtsn: principal separability; Boruta, RandomForest, mixOmics for classification, and correlation approaches. For comparison, Sparse Partial Least Squares for Discriminant Analysis was used.


Within the MARIA analysis workflow, we show, that, A) in the initial analysis step featuring tSNE, CD patient data can be separated from the control subjects based on the high dimensional multimodal MR dataspace. Next, in terms of classification B) we achieved the best balanced accuracy in separating the two groups with the resting-state graph info parameters, which characterize the whole resting-state network by graph-theoretical parameters like small world index, density (Fig. 1), followed by anatomical freesurfer data (Fig. 1) and lastly by node specific resting-state data (not shown). Moreover, C) in all cases we got a quite high balanced accuracy which was D) always higher for sPLS-DA (Fig. 1). Finally, cross-correlation approaches also provided a separation of CD and controls and further hint to brain structure specificity. As an immediate next step, the best available bio-signature will be generated incorporating all multimodal data available and correlate this to clinical data.


MARIA, a GUI based analysis tool, integrates 1) assessing a principal separability (tSNE) 2) extracting bio-markers separating the experimental groups (random forest / sPLS-DA) within each imaging modality but also across modalities. Finally, these imaging bio-markers can be flexible correlated to clinical data leading to hypothesis generation but also direct hypothesis testing. For CD – control separation resting-state is better than anatomy.

AcknowledgmentWe thank Dr. Pfleger Stiftung and the IOIBD for financial support of this study as well as all patients and control subjects.
[1] Hess et al., 2015. Functional Brain Imaging Reveals Rapid Blockade of Abdominal Pain Response Upon Anti-TNF Therapy in Crohn's Disease., Gastroenterology 2015, 149(4):864-6, doi: 10.1053/j.gastro.2015.05.063

[2] Reuter, M., Schmansky, N.J., Rosas, H.D., Fischl, B. 2012. Within-Subject Template Estimation for Unbiased Longitudinal Image Analysis. Neuroimage 61 (4), 1402-1418. http://surfer.nmr.mgh.harvard.edu
[3] Kreitz et al., 2018. A New Analysis of Resting State Connectivity and Graph Theory Reveals Distinctive Short-Term Modulations due to Whisker Stimulation in Rats. Front Neurosci. 2018 May 23;12:334.
Fig. 2 Crosscorrelation matrices of anatomical and resting state data of CED and controls
MARIA snapshots depicting different brain structure driven cross-correlation matrices based on upper row: e.g.  volume from freesurfer analysis and lower row: MRSA resting-state analysis, here hub-score of each node comparing CED patients left colum from controls right column.
Fig. 1 Classification of CED vs. controls on free surfer anatomical data and resting state data

MARIA snapshots depicting bio-marker derived from the aseg freesurfer atlas and corresponding geometric properties separating CD from controls in chord-plots (upper row) and the resting state graph-info parameters (lower row) for sPLS-DA (left column) vs. Boruta random forest (right column). Resting state data lead to better balanced accuracy for separating CD patients from controls compared to anatomical data. In both cases sPLS-DA leads to superior separating accuracy but it is quite high  for both classification algorithms.

Keywords: Crohn's disease, multimodal imaging, resting state, fMRI, tSNE