Skip to main content

Hybrid Fluorescence Molecular Tomography and X-ray Computed Tomography system and method

Final Report Summary - FMTXCT (Hybrid fluorescence molecular tomography and X-ray computed tomography system and method)

Executive summary:

As overall goal of this project a fully functional FMT-XCT prototype, based on a commercial XCT system, has been developed. In-vivo studies have been performed using proprietary, user-friendly software to control the acquisition. Results were published in high ranking journals e.g. Nature methods Ale A. et al., 2012 'FMT-XCT: in vivo animal studies with hybrid fluorescence molecular tomography-X-ray computed tomography'. A self-developed gantry which could offer fast acquisition times through minimization of FMT/XCT interference was developed by CT Imaging small and medium-sized entreprises (SME). As intended CEA Leti has designed and implemented a dual energy micro CT imaging system. Both the functional prototype and calibrated, dual energy processing software have been demonstrated during the training session and are available to the consortium.

We engineered the optimal theory and inversion approaches to achieve the best performing synergistic system. FORTH in strong collaboration with UZH, UCL, CEA-LIME and FIHGM/UC3M has developed an ultra-fast inversion algorithm which was extensively tested with phantom and experimental data. A multispectral algorithm and a user friendly software, dramatically improving usability, efficiency and attractiveness have been developed for the end user. Furthermore a fast image reconstruction algorithm using data compression and a combined reconstruction classification method for diffuse optical tomography were developed and published. As intended in the proposal we successfully developed and evaluated various cancer animal models as well as different targeted fluorescent probes (Cea Lime) which are available to all partners. Furthermore essential aspects of hypoxia induced signaling were investigated. Partner 6 (UZH) developed, financed by another source, a hybrid FMT-MRI system. To evaluate the imaging performance of the FMT-XCT scanner and compare with PET/CT, different optical phantoms were customized and distributed amongst the partner. Furthermore a new methodology was implemented to allow the co-registration of FMT-XCT and PET XCT images. Both imaging modalities were validated for a specific application, in-vivo using dual-probes and compared in terms of quantitative accuracy and performance metrics. In addition to regular consortium meetings four training session at different partner locations (Grenoble, London, Zürich, Munich) were carried out.

Conclusive note: Our team impressively completed all the technical goals. Results were published in high ranking journals (see final report for the complete list of publication). The collaboration was excellent; there were no known problems within the network.

Project Context and Objectives:

The overall goal of this project, is to develop a truly unique imaging system, the likes of which exists nowhere by combining Fluorescence Molecular Tomography (FMT) and X-ray CT (XCT) into a hybrid, quantitative system and method, engineer the optimal theory and inversion approaches for achieving a highly performing and synergistic system and perform pre-clinical imaging with a view towards clinical translation and therapeutic intervention. The work of the project was split into nine work packages including two work packages (WP1 and WP9) exclusively dealing with management, co-ordination, training and dissemination activities.

Project Results:

As overall goal of this project a fully functional FMT-XCT prototype, based on a commercial XCT system, has been developed and in-vivo studies have been performed. Our team impressively completed all the technical goals. Results were published in high ranking journals (see final report for the complete list of publication).The collaboration was excellent, there were no known problems within the network.

Please find below a summary of the main results split by work package.

WP 1: Management (WP leader HMGU)
The management structure of the project was as follows: The Helmholtz Center Munich (HMGU) as a coordinator (scientific coordination: Prof: Dr. Vasilis Ntziachristos, administrative coordination: Dr. Christiane Ogorek, Dr. Veronika Erben) was directly responsible for reporting to the European Commission. The members of the executive committee are the work-package leaders in this proposal (besides partner CT Imaging). This structure allowed a direct communication between partners and a common understanding of goals. Each member of the executive committee further directly supervised or steered his team, and has selected members that further allow the direct communication within work-packages. The advisory committee consisted of senior external members that are well respected in scientific fields associated with this proposal. The Executive committee was interacting with Advisory Committee members on the basis of personal correspondence and private meetings at the site of the Advisory member or in international meetings. Different experienced management teams provided support in financial, legal and PR issues.

Coordination and Communication activities
The communication was working excellent- internally and externally. Additional to the project correspondence and the day-to-day requests from partners the coordination and communication included (detailed description see periodic reports 1-4):

-Reporting to the European Commission.
-Preparation and submission of an amendment request
-Financial and scientific reporting
-Preparing and post-processing of project meetings
-Update of the project website
-Preparing press releases

WP 2: XCT Development (WP leader Cea -Leti)
The global objective of WP2 was to provide an X-ray Computed tomography (XCT) module that can be integrated with a Fluorescence Molecular Tomography (FMT) module into a hybrid imaging system. Several designs and commercially available systems exist for X-ray CT of small animals.

The need to develop a specific X-ray CT system comes from the diverse needs of the hybrid approach to produce an XCT design that is not only appropriate for small animal imaging but also:
-provides adequate accommodation of the optical components,
-eliminates X-ray interference with optical components,
-offers improved contrast between organs as is important for the optimal utilization of X-ray CT information as priors in the FMT inversion procedure, as explained and performed in WP4.
The milestone M5, 'Selection of an appropriate XCT technology for FMT-XCT system' finalizes WP2, for which 7 deliverables were issued (D2.1 to D2.7).

The XCT prototype and the implementation of a dual-energy X-Ray acquisition protocol has been developed based on LETI bench and simulation (D2.1 2.2 2.3 and 2.4) validated by experiments on various phantoms and living mice. The contrast between organs and tissues achieved by the dual energy method has been evaluated and compared to those achieved by contrast agents (D2.6) with the help of FIHGM. The minimization of interference of X-ray with optical components has been considered (D2.5). Complete specifications have been issued (D2.7) by CT-Imaging.

These works lead to two conclusions:
-it is easy to distinguish between bones and soft tissues, feasible to distinguish between adipose and other soft tissues, but almost impossible to identify organs,
-high energy does not add significant information (unless very high dose are used both energies).

Consequently we suggest performing only the low energy acquisition, the second full tomographic acquisition being not justified.

Additional experimentations have been performed on the final prototype in Munich in May 2011 (part of WP5) and confirmed these conclusions. Contrast evaluation showed a little improvement with respect to what was obtained on the CEA-LETI bench thanks to the fact that on the FMT-XCT machine a much lower inherent filtration is present, which permits to exploit also the information carried of the lowest part of the emitted x-ray spectrum much better than what could be done on the CEA-LETI scanner.

WP 3: Theory for 360-degree FMT (WP leader FORTH)
During the first year of the project FORTH's involvement has been mainly towards developing new algorithms for fast and more efficient inversion as well as multispectral reconstructions for multiple target detection. We have developed and tested a direct inversion method based on calculating fluorescence in Fourier space. This algorithm has proved significantly better than the standard algorithms when a large number of sources are used but is not convenient for FMT acquisitions where usually a small number of sources are used. Furthermore, we have developed a new algorithm for boundary removal of arbitrary surfaces. The multispectral method is based on a linear unmixing algorithm and has been tested with both phantoms and in vivo studies and is capable of resolving different fluorescing and absorbing targets. At the same time a very large number of experimental data sets has been produced and made available to all partners of the consortium for algorithm comparison, optimization and finalization. Finally, a user friendly software has been developed that can be incorporated to all FMT systems and dramatically improves usability, efficiency and attractiveness to the end user. It can be used for data analysis and visualization of results in a fast and efficient way.

During the second year the main work that has been performed by FORTH has been:
-The development of matrix-free inversion approach for Finite Element Methods (ultra-fast inversion).
-The development of software to account for source profile, which will be crucial for correctly implementing a 360 FMT machine.
-The development of user friendly software for data inversion and running experiments and the comparison of newly developed codes with pre-existing standards such as Finite Element Methods (FEM).

The work performed during the 3rd year of the project has been focused on:
-Finalizing the multispectral algorithm, implementing the multispectral capacity into the fast inversion approach and quantitatively evaluating the direct inversion with hybrid data.

During the last year of the project FORTH has worked in closed collaboration with HMGU and UCL and produced results within Work Package 4: 'FMT inversion with image priors'. These work included evaluating priors with the Fast Matrix Free Method and developing the visualization tool for the reconstruction as well as complete sensitivity estimation.

Main and very strong collaboration with exchange of data and personnel has taken place during this reporting period with ETH, UCL, CEA-LIME, HMGU and FIHGM.

WP 4: FMT inversion with image priors (WP leader UCL)
Due to the diffusive nature of light propagation in biological tissue the image reconstruction in fluorescence molecular tomography (FMT) is an ill-posed inverse problem, which leads to relatively poor spatial resolution and reconstructions that are highly sensitive to noise. Furthermore, the inhomogeneous optical properties of the animal make the image reconstruction even more challenging because of the difficulties in modeling the forward problem accurately. Another challenge of FMT is the reconstruction of images from large data sets, since it increases the computational time and complexity.

We have developed methods for overcoming the challenges described above and we are now able to recover the fluorophore location, size and concentration with higher accuracy than conventional methods.

Improvements in resolution and overall image quality were achieved by incorporating tissue structural information derived from X-ray computed tomography (CT) images into the FMT image reconstruction problem. We introduced a split operator method to reconstruct fluorescence images fast and efficiently, using a nonlinear anisotropic diffusion regularization term that incorporates anatomical prior information.

In order to reduce the dimensionality and complexity of the problem, we developed data and solution compression methods, which provide faster reconstructions without significant loss of information.

We developed methods that combine an estimate of the optical properties of the tissue and classification/segmentation techniques, which are used to improve the accuracy of the forward model, and hence, the accuracy of the solution.

WP 5: FMT-XCT integration (WP leader HMGU)

One of the main tasks of the project was to install the final FMTXCT at HMGU, which was finalized in cooperation with Ct-imaging in year 3 of the project (details see periodic report 3). Algorithms for the assignment of optical properties were integrated in the reconstruction code and a user friendly software has been installed and is operational on acquisition.

During the last reporting period focus was on in vivo experiments as described in the periodic report 4 (see also WP6 and 7).

Finalization of the FMTXCT prototype
The proprietary gantry prototype was finalized in cooperation with Ct-imaging. Thus to accelerate the system integration of both the FMTchain and the CT hain HMGU and CT-Imaging worked out a concept to provide a fully functional and operational prototype. CT Imaging was commissioned by HMGU to realize the concept. The concept comprised of basically four steps:

-Providing a mechanical interface for the FMT components
-Upgrading the machine control to support some FMT components
-Providing a software tool enabling the user to start a standard CT acquisition
-Providing an interface between the machine control and the existent FMT- application used by HMGU.

All tasks have been successfully accomplished by CT-Imaging and the system was delivered to HMGU (Klinikum Rechts der Isar, Munich) on March 09, 2011. (for details see deliverable 5.4).

Integration of algorithms from Wp3 and Wp4
One of the steps in the integration of algorithms was the development of a common file format. The developed common file format for exchange is called fmtxml file format. The format has been agreed upon and used by all cooperation partners. In addition to the file format a matlab reference implementation for importing and reading the dedicated files was developed and was published to the partners. Several datasets that have been uploaded to the transfer drive. All partners were able to download the data from the server and the datasets have been used to evaluate the developed components of the inversion methods. The ultra-fast inversion method has been successfully applied. A user friendly software was installed on the acquisition computer, see deliverable 5.9. The user can choose optimal parameters for acquisition; see task 5.4 and deliverable 5.8.

WP 6: Cancer Imaging (WP leader CEA-lime)
Workpackage 6 concerned the development of fluorescent probes and models for the in-vivo evaluation of the FMT-XCT system. Emphasis was given to the breast cancer imaging since breast cancer represents a common cancer type and it is one of the few cases that FMT-XCT can be implemented in clinic. In the course of the project, several cancer models were evaluated for FMT imaging. More specifically, mice bearing xenografts from human breast cancer cell lines (MDA-MD 231 and the MCF-7) as well as a transgenic strain of mice (PymT) that develop spontaneous breast cancer. We also performed some experiments in a xenograft model of glioma (U87 cell line) to test FMT imaging on another cancer model. For the case of the fluorescent probes, we used several commercial probes and also custom made probes. The commercial probes concerned the Integrisense and Prosense probes from PerkinElmer and Angiostamp from Fluoptics. For the case of the custom probes, we have reported a new aptamer,ACE8 capable of binding the annexin A2 protein of MCF7 cancer cells with high affinity. Experiments on the above models have been carried out in CEA-LIME. Furthermore, MDA-MB 231 xenografted mice and the transcenic model of PymT have been shipped to HGMU in Munich for imaging with the FMT-XCT system. One of the main objectives of the FMT-XCT program was to validate the quantification capacity of FMT. For this reason, the FMT system TomoFluo3D was compared to a highly quantitative and well establishing technique such as PET imaging. Such experiment allowed us to prove, that FMT is capable of achieving sensitivies in the order of 1 pmole and its linearity spans over three orders of magnitude.

Our other major task was to use fluorescent probes for imaging cancer. Our approach involved co-registering imaging of PET with imaging of FMT for enhancing the information that can be extracted from a single subject. Two applications of this approach were explored:
a) the use of PET as a reference in order to evaluate new fluorescent probes that are introduced as PET probe analogs and
(b) complementary imaging in which optical probes and nuclear probes have target separate processes.

In the first case a fluorescent glucose analogue, the 2-DG IrDye800 was found to have a different distribution with respect to the reference probe of tumor imaging, the nuclear probe 18F-FDG . In the second case cathepsin activity and integrin avb3 distribution was imaged with respect to the 18F-FDG imaging.

During the whole project runtime a plurality of different cancerogenous organs where invenstigated successfully using our FMT-XCT imaging scanners and reconstruction algorithms. For all aquired imaging studies validation was performed by a novel fluorescent and visible imaging volumetric cryoslicing technique. Performed Imaging include but are not limited to subcutaneously implanted mammary carcinoma (4T1) neck tumors, osteogensis impferfecta mouse model for visualizing bone remodelling and a Lung tumor study (Kras+/- and BL6 Tyr-/-).
The results were published in Ale A. et al. (2012). FMT-XCT: in vivo animal studies with hybrid fluorescence molecular tomography-X-ray computed tomography, Nature Methods, 9, 615-620, (2012), doi:10.1038/nmeth.2014

WP 7: Imaging cancer therapy (WP leader UZH)
-Technology development (financed by ETH):
We have developed and characterized a prototype FMT-MRI system, which operates at a magnetic field strength of 9.4 T. The in-plane spatial resolution achieved was 1.5 to 2mm, while depth resolution of the current setup is 3mm, limited by the restricted field-of-view. Dose linearity has been demonstrated by correlating fluorescence intensity to the amount of dye (AOI987) in the sensitive volume, the sensitivity limit was estimate to be 0.3 picomole of the dye. First proof-of-principle studies have been completed in tumor bearing mice, using FMT to assess protease activity and MRI to provide the anatomical correlate.

In a second phase an improved version of this prototype has been built with a CMOS detector replacing the SPAD and with improved mechanical stability. The new system had an improved technical performance; the principal drawback was the detector sensitivity, which was found to be one to two orders of magnitude inferior to a high-performance cooled CCD camera. We are currently working on a third prototype, which will comprise a camera of improved sensitivity.

-Development of assays for fluorescence imaging:
i) Hypoxia signalling: The multimodality imaging tool allowing the assessment of various steps in the HIF signalling cascade [4] has been extended to allow for fluorescence imaging of hypoxia inducible factor (HIF) activity. Two approaches have been pursuit to monitor HIF activity. This was achieved by expressing different reporters under the control of the HRE (HIF responsive element) motif comprised in the promoter sequence of HIF target genes. In a direct translation of the bioluminescence assay we was either by replacing the firefly luciferase reporter by a fluorescence reporter (mCherry) or by using a reporter construct that was expressed at the cell surface and could be targeted by exogenously administered fluorescence probes. Activation of the HIF signalling cascade led to the expression of both reporter constructs which could be directly demonstrated by observing the mCherry fluorescence or following the administration of a biotin based imaging probes labelled with Alexa680. The elegance of latter assay is that by replacing the label, e.g. using a radioligand as reporter, different imaging modalities could be used. This is relevant when validating the FMT readout. In the meantime a number of tumor cells have been transfected with either of the two constructs (C51 colon carcinoma, 4T1 mammary carcinoma, GL261 glioma).
ii) Antibody based targeting of pancreatic tumor cells: We have generated a monoclonal antibody 8/9-mAb, targeting TMEM27, a surface N-glycoprotein highly expressed on beta-cells, compared its expression in human and mice pancreas and demonstrated beta-cell specific binding in both humans and mice. In vivo imaging was performed in mice with subcutaneous hTMEM27 over-expressing insulinomas using fluorescently and radioactively labeled antibody, followed by tissue ex vivo analysis and fluorescence microscopy. In vivo fluorescence imaging in nude mice with insulinoma xenografts expressing hTMEM27 showed high 8/9-mAb uptake in tumors after 72h. An issue with antibody probes is slow systemic clearance, therefore optimal signal-to-background rations have been obtained only after 72h. FMT was required for derivation of quantitative information on TMEM27 expression levels.

-Application of imaging tools to assess treatment response in experimental cancer
An unexpected observation during the assay development phase was that persistent tumor hypoxia led to a decrease in HIF signaling intensity. This was also confirmed by in vitro analysis, which revealed a partial disconnect between areas displaying hypoxia and areas showing HIF1- stabilization and HIF downstream activity. Obviously the clinical established PET readout of hypoxia does not represent the response of the biological system to hypoxic conditions.

The 2-oxoglutarate dependent prolylhydroxylases (PHDs) are key regulators of cancer progression beyond their role in cellular oxygen sensing. While PHD inhibitors are currently being evaluated for the treatment of renal anemia in clinical studies, their potential application for the treatment of cancer has not yet been tested. To investigate the effect of chronic, non-selective inhibition of proly-4-hydroxylases on cancer progression, we treated U87 and C51 cells with the PHD inhibitor dimethyloxalylglycine (DMOG). We have used the imaging methods described in combination with positron emission tomography (PET) and magnetic resonance imaging (MRI)) to in vivo monitor HIF transcriptional activity, tumor oxygenation, vascularization and metabolic activity in s.c implanted tumors of mice chronically treated with this drug. Strikingly, we wound that long-term administration of DMOG resulted in a decrease in HIF transcriptional activity in these tumor allografts, whilst there was no significant effect on tumor growth, oxygenation level, tumor blood volume and vessel density. MRI and PET experiments demonstrated that there was no significant global effect of drug treatment on the tumor microenvironment (angiogenesis, glucose metabolism). Yet, advanced analysis of tumor heterogeneity revealed significant differences in the vascularity. In vitro experiments in tumor cells demonstrated that DMOG induces a negative feedback loop which acts to decrease HIF1 protein levels thereby down-regulating HIF transcriptional activity. The molecular players involved in this feedback loop remain to be elucidated.

-FMTXCT in-vivo experiments
In-vivo experiments were performed at HMGU to demonstrate that the FMTXCT prototype can depict different stages of tumor progression. The corresponding paper is still under review. Details are explained in deliverable 7.5 and 7.6. (Milestone 11).


Lead by: FIHGM / UC3M
The main objective of this workpackage was to validate the hypotheses driving this proposal, i.e. that an FMT-XCT system can result in imaging performance similar to the obtained with nuclear medicine imaging, and more specific PET-XCT. Two sub-objectives were defined to achieve this goal. The first one was to validate FMT-XCT with PET-XCT for a specific in-vivo application using dual-probes. This required designing a methodology that allowed accurate co-registration between FMT-XCT and PET-XCT modalities. This was done by building a mouse holder that constrains the mouse movement when the mouse was imaged in different systems. Thus, images of the three modalities were accurately and easily co-registered by image alignment using a custom made, user-friendly software application. Then the two functional modalities (FMT and PET) were compared in terms of quantitative accuracy. The second sub-objective was to define a protocol for assessing FMT and PET in terms of performance metrics. For that purposed we started from the NEMA-NU4 standard applied to small animal PET. Then we defined an 'equivalent' protocol applicable to FMT and used it to assess the FMT to PET comparison. Results indicate that FMT requires a more elaborate quantification protocol in order to obtain results comparable between different imaging systems. We have found quantification accuracy similar in PET and FMT and provide guidelines to establish a protocol for FMT imaging.

WP 9: Training and Dissemination (WP leader HMGU)
As described in Annex 1 the proposal depended on large extend on discussion and exchange of ideas and interdisciplinary scientific knowledge between participant members. Therefore, as intended in Annex 1, several workshops have been realized:
-Workshop 'Advanced X-Rays imaging techniques', 6-7 July, 2009, Grenoble organized by CEA Leti (for details see report 2, WP9 and deliverable 9.2).
-Workshop 'Reconstruction Methods', 16-18 September, 2009, London, organized by UCL (for details report 2, WP9 and deliverable 9.2).
-Workshop 'Animal models/cancer imaging', Zurich January 13, 14th, 2011 (for details see report 3, WP9 ).
-'Workshop on FMT XCT' at HMGU, Munich November 23-25th, 2011 (for details see Deliverable 9.5)

The objective of dissemination activities was to deliver relevant project results to key target groups and to improve the relevance of results by a continuous dialogue with these stakeholders. Moreover visibility and awareness of the project are enhanced through such activities and thereby decision making is influenced.

The results of the FMTXCT project were disseminated in various ways:
-Publication in leading international journals in the field and annual presentation of results at international forums of Imaging (details see list of publication A1 of the Final Report)
-Dissemination of the FMTXCT project and the results were additionally obtained through the project website (see online) and several other websites (details see deliverable 9.1 'Dissemination and Implementation document' and table A2 of the Final Report).
-At the conference 'Scientific Challenges in European Health', an international conference launched by the Bavarian Universities, which took place in Brussels on October 20 and 21, 2010 Vasilis Ntziachristos was selected to present the Bioimaging topic. The objective of the event was to present major themes of medical research in Bavaria and scientific achievements in these fields as well as to discuss key challenges, goals and new directions with international stakeholders from science, industry and European institutions. In the imaging presentation and the following debates one of the main focuses was the FMTXCT project (Poster see deliverable 9.1 'Dissemination and Implementation document')
-A public promotion leaflet to raise awareness for the activity and success of the project has been developed (see deliverable 9.3).
-The following press release (see online) was prepared and distributed via IDW ('Informationsdienst Wissenschaft': see online)

Potential Impact:

Medical imaging has altered the processes of diagnosing, treating and studying diseases. Optical imaging strategies are emerging tools to facilitate the early detection and diagnosis of diseases. The imaging system developed within this project advances multi-modality imaging by offering the first FMT-XCT system worldwide and thereby increases the competitiveness of European health care biotechnology and medical technology sectors. By combining X-ray computed tomography (XCT) and fluorescence molecular tomography (FMT) the system generates a 360° image and provided much better results in living mice than with either method on its own. The combined use of two proven procedures opens up great possibilities for non-invasive diagnostics, for example in lung cancer. Due to this new development it is possible to diagnose much more precisely where tissue changes have taken place. In the future the FMT-XCT system will be further refined so that it can also be used in clinical diagnosis of humans. Consequently patients would benefit enormously since FMTXCT imaging would allow a more efficient diagnosis by precisely locating diseased tissue and additionally is noninvasive. The increased use of imaging in clinical imaging would also save money due to improvements in the efficiency of diagnosis and treatment.

List of Websites: