PARAMAP: an Automated Imaging Analysis Tool for Quantitative CEST Molecular Imaging: Validation in vitro J. Flament 1 , B. Marty 1 , S. Mériaux 1 , J. Valette 1 , C. Medina 2 , C. Robic 2 , M. Port 2 , F. Lethimonnier 1 , and F. Boumezbeur 1 1 NeuroSpin, I2BM, Commissariat à l'Energie Atomique, Gif-sur-Yvette, France, 2 Research Division, Guerbet, Roissy-Charles de Gaulle, France Introduction Recently, a new class of paramagnetic contrast agent has been developed for Chemical Exchange Saturation Transfer (PARACEST) magnetic resonance imaging [1-3]. Since visualizing CEST contrast requires two measurements with B 1 saturation applied on-resonance (at + , frequency of the shifted bound water) and off-resonance (at - ), CEST imaging is sensitive to inhomogeneities in both B 0 and B 1 fields. Therefore, in order to generate quantitative CEST maps, it is important to elaborate correction algorithms to get rid of errors induced by B 0 and B 1 fields. In this study, we proposed to use a numerical simulation of the CEST contrast mechanism based on the Bloch equations modified for chemical exchange incorporating B 0 and B 1 dependencies [4]. The efficiency of our analysis tool was verified in vitro. Materials and Methods MRI acquisition. Experiments were realized on a 7 T small animal MRI scanner (Bruker, Ettlingen, Germany) using a bird-cage 3-cm-diameter 1 H coil for acquisition and reception. CEST images were acquired with a RARE sequence (TE/TR=80/5500 ms; turbo factor 32) preceded by a CW saturation pulse being applied at ± 50ppm (T sat =400ms, B 1sat ~20μT). B 0 and B 1 maps were acquired separately using a GE sequence (TE=5, 7.5, 10, 15ms; TR=300ms, flip angle of 30° and 60°). In vitro tests were performed on a 6-tubes phantom each containing [Eu 3+ ]DOTAM-Gly (Guerbet, Roissy, France; concentrations of 0.5, 1, 2.5, 5, 7.5, 10 mM) [3] embedded in a low-gelling point 4% agarose matrix. Z-spectra Simulation and Image Analysis with PARAMAP. Our image analysis tool designed as PARAMAP is a Matlab (The MathWorks Inc., Natick, MA) based program aiming at correcting the B 0 and B 1 induced errors on the native CEST image (I CEST =(I ON -I OFF )/I REF ). Briefly, PARAMAP simulates for each pixel r a series of asymmetric Z-spectra using B 1 (r) and B 0 (r) values with the concentration C as a variable (aMTR(C,r)). The others parameters of the simulation (k ex , , T 1 and T 2 of both pools) are extracted from experimental Z-spectra of [Eu 3+ ]DOTAM-Gly (data not shown). The concentration map C(r) is then calculated from the minimization of the cost function: |I CEST (r)-aMTR(C,r)|. Results and Discussion As illustrated by the figure 1, field inhomogeneities manifest themselves strongly on the amplitude of the observed CEST effect for a given concentration. Therefore a 10% error on B 1sat leads to a 4% over- or under-estimation. Similarly, a 100Hz frequency error leads to a 1% over- or under-estimation. In our experiment, B 0 et B 1 inhomogeneities were quite modest as illustrated (standard deviations: B0 =21Hz and B1 =0.5μT), yet without correction, the calculated %CEST effect (Fig.2, open red dots) is quite different to the %CEST effect expected (blue dots). If not corrected, discrepancies between the known and the estimated concentrations are on average of 0.8mM. The B 0 and B 1 corrections (green line) improve significantly the quantitativity of the established PARACEST concentration map with an averaged over-estimation of 0.3 mM (See Fig.3). Conclusion CEST agents are promising new contrast agents for MR molecular imaging since they allow to reach nanomolar sensitivity [5]. Yet, their susceptibility to parameters such as B 0 , B 1 is a real issue to achieve truly quantitative CEST imaging. In this study, we validated in vitro PARAMAP, a home-made software aimed at correcting not only for B 0 and B 1 field inhomogeneities. Ultimately, quantitative PARACEST concentration maps were obtained within a reasonable margin. To move further toward in vivo quantitative CEST imaging, we are actually extending the simulation to a 4-site chemical exchange model similar to the one described by Li et al. [6]. The software will be available at: http://groups.google.com/group/paramap . Acknowledgments Grant sponsor: Iseult/Inumac French-German Project. References 1. Ward KM et al., J Magn Reson 2000;143:79 2. Zhang S et al., Acc Chem Res. 2003;36:783 3. Aime S et al., MRM 2002;47:639 4. Woessner DE et al., MRM 2005;53:790 5. Terreno E. et al., CMMI 2008;3:38 6. Li AX et al., MRM 2008;60:1197 Fig.1. 3D asymmetrical Z-spectrum for [Eu 3+ ]DOTAMGly exhibiting variation of the %CEST effect for different B 1 intensities and saturation offset frequencies. Fig.2. Comparison of %CEST effects within each tube before (open red dots) and after correction for B 0 - and B 1 - induced errors (open green dots) to the expected %CEST effect (blue dots). The errors are reported in the box at the top-left corner. 20 18 16 (in Hz) (in μT) (in mM) 30 0 -30 60 20 18 16 (in Hz) (in μT) (in mM) 30 0 -30 60 Fig. 3. Color coded quantitative PARACEST concentration (C) map after corrections for B 0 and B 1 inhomogeneities (top left corner and top right corner). The mean concentrations calculated in each tube are given explicitly (in mM). Proc. Intl. Soc. Mag. Reson. Med. 17 (2009) 4652