1. WP3: Theory for 360 degree FMT Progress during Year 1 Giannis Zacharakis Institute for Electronic Structure and Laser (IESL) Foundation for Research and Technology – Hellas (FORTH) FMT-XCT First Annual Meeting April 24 2009

2. Whole animal imaging Whole Animal In Vivo Optical Imaging Group IESL – FORTH Cellular and Sub-Cellular

3. Ballistic Propagation (OPT) Ballistic Propagation Beer´s Law ~ exp (-a*z)

4. FMT Highly Scattering Fluorescence Molecular Tomography Diffusion ~ exp (-a*r)/r

5. 2 nd generation FMT system

6. Inverse problem: F (r) = f -1 [X inc ; X prop ] Forward problem: X prop = f [X inc ; F (r) ] Tomographic imaging Iterative Methods (ART), SVD Normalized Born Approximation + Homogeneous Medium F

7. transgenic age ~ 8 weeks non-transgenic age ~ 8 weeks Normalized measurements hCd2-GFP mouse GFP Control

8. FORTH’s participation Objectives - 2.2.3.1 To implement direct inversion based on boundary removal method for media with arbitrary boundaries - 2.2.3.2 To research optimal direct inversion approach with simulations and experimental data - 2.2.3.3 To compare the direct inversion performance with conventional, previously developed FMT inversion methods - 2.2.3.4. To incorporate algorithms for multi-spectral imaging - 2.2.3.5. To develop user friendly software for inversion of XCT-FMT data based on direct inversion approaches - 2.2.3.6. To invert training data acquired from FMT-XCT system for algorithmic finalization

9. Objectives 2.2.3.1 To implement direct inversion based on boundary removal method for media with arbitrary boundaries - 2.2.3.2 To research optimal direct inversion approach with simulations and experimental data - 2.2.3.3 To compare the direct inversion performance with conventional, previously developed FMT inversion methods Theory for 360 FMT

10. Boundary removal Ripoll and Ntziachristos, PRL 2006.

11. Assuming the interface locally plane at (R,z0), we can propagate the measurement U to anywhere in (diffuse) space: (first Raleigh-Sommerfield integral formula) (R,z0) (R,z) (infinite Diffuse Medium - virtual) Original Diffuse Volume Boundary removal - Use of infinite Green’s functions - Apply direct inversion 1.Backpropagation 2.Complete Fourier Approach

12. Complete Fourier Approach Direct Inversion! Problem in Real space: Problem in Fourier space:

13. LIMITATIONS: 1) It can only be used with large numbers of sources Grids of 64x64 sources to obtain relatively good data and are still prone to great Fourier artifacts. This is a great problem since all our FMT setups work with numbers of sources in the 100 range. Having more sources in unpractical, since the experiment times will increase unnecessarily. 2) Fluorophore concentration in Fourier space. This means that once inverted the data has to be Inverse-Fourier transformed in 3D. This yields significant Fourier artifacts (seen as 'waves' surrounding the main central points of data) that worsen as the number of sources or detectors gets smaller. This means that in practice even larger numbers of sources need to be used. Complete Fourier Approach

14. After developing and implementing the main features of the direct inversion method, we have decided it is not the method to pursue for our FMT setups. Clearly we need a method that can solve fast the inversion needed but can work with small numbers of sources, in the order of <100 sources. We therefore opted to change this deliverable, DIRECT INVERSION method to a deliverable called 'ULTRA FAST INVERSION for FMT' that will be delivered on the next reporting period. In the meantime, the partners have access to an inversion method which is significantly faster that the currently existing ones, based on: Boundary removal and ART inversion on the infinite homogeneous data. Conclusion

15. Objective 2.2.3.4. To incorporate algorithms for multi-spectral imaging Multi-color imaging - A multispectral algorithm has been developed and tested for simultaneous detection of multiple fluorophores and absorbers. - It will be incorporated in the FMT software in the next reporting period according to the time schedule.

16. Two or more fluorophores with overlapping emissions Detect fluorescence in equal number of channels  In microscopy it is performed pixel-by-pixel on the so-called spectral cube In tomography we must perform it voxel-by-voxel on the Reconstructed Data Multi-color imaging

17. Tomographic calculations Fluorescence reconstruction Absorption reconstruction In the case of two fluorophores

18. Slab phantom: Intralipid + ink + Agar μ s = 16 cm -1, μ a = 1.5 cm -1 Phantom study

19. Multicolor Quantification study CFSE: 4μM ATTO590: 5μM, 10μM and 15μM Recover the correct concentrations only when applied on reconstructions

20. In vivo tissue phantom CFSE: 4μM ATTO590: 5μM, 10μM and 15μM

21. Tissue oxygenation Use fluorec to segment the oxyrec data

22. Image concurrently in 3D - Fluorescence activity - Oxygenation in hypoxic regions - Neo-vascular factors related to tumor proliferation - Measure dynamic parameters (BV and OxySat) - Identify cancer stage and phenotype (benign or malignant) Tissue oxygenation

23. Objective 2.2.3.5. To develop user friendly software for inversion of XCT-FMT data based on direct inversion approaches Software development Custom developed software based on Labview suitable for: 1.Data acquisition 2.Data analysis (OPT and FMT) Automated user-friendly software for data analysis Uses open source application for visualization e.g. ImageJ, Amide, Osirix

24. Software development

26. Mini-FMT Automated software for FMT acquisition and analysis The main goal of the FMT software is to provide a user-friendly interface to take the measurement data and later perform the 3D reconstruction providing an image that can be analyzed with Open Source applications. e.g. ImageJ, Amide, Osirix There is a complete manual available to the partners with detailed information and guidance for all the functions and parameters. One button reconstruction

27. Mini-FMT

28. FMT data to be shared Objective 2.2.3.6. To invert training data acquired from FMT-XCT system for algorithmic finalization A large number of experimental measurements have been acquired that are available to all partners for optimization and finalization of algorithms. These measurements involve phantoms as well as in vivo experiments: Quantification Resolution Multispectral In vivo studies (lymph nodes, tumor progression, oxygenation)

29. FMT data to be shared

32. FMT – TOAST comparison Exchange of data with Partner 4UCL Reconstructions by: Athanasios Zacharopoulos

33. FORTH – LIME data Experiments performed with FORTH-FMT for comparison with LIME-FMT Visit of Anikitos Garofalakis to FORTH

34. FORTH – LIME data

36. Second Year Ultra fast inversion for FMT (new Deliverable) Test and compare the performance Optimize and finalize Mini-FMT (feedback from Partners) Incorporate the multispectral algorithm (absorption & fluorescence) Increase the exchange of data! (essential for joint progress)

37. Vasilis Ntziachristos Funding E.U. Integrated Project - “Molecular Imaging” E.U. STREP - “TRANS-REG” E.U. EST – MOLEC IMAG E.U. Collaborative Project – “FMT-XCT” Dimitris Kioussis IN-VIVO IMAGING GROUP Jorge Ripoll Giannis Zacharakis (Post-doc) Ana Sarasa (Post-doc) Udo Birk (Post-doc) Rosy Favicchio (PhD student) Alex Darell (PhD student) Maria Simantiraki (Msc) FORTH – IESL/IMBB: E. N. Economou Clio Mamalaki Sifis Papamatheakis Nektarios Tavernarakis Past members Juan Aguirre Abraham Martin Anikitos Garofalakis Heiko Meyer Stelios Psycharakis Sascha Atrops (EST trainee) Olga Kravtsenyuk (Post-doc) Lucie Lambert (EST trainee) In vivo Imaging Group Simon Arridge Athanasios Zacharopoulos Bertrand TAvitian Anikitos Garofalakis LIME - CEA Collaborations

39. Geometry Inversion: Use spectra Wavelength dependent W Depth dependent W Spectral unmixing On going

40. CFSE (20μM) ATTO590 (20μM) In vivo tissue phantom

41. Same number of GFP & DsRed T-cells injected Only DsRed cells recognize an antigen peptide Repetitive imaging of cervical lymph nodes Following T-cell numbers