The Purkinje conduction network plays a crucial role in normal cardiac function but has also been implicated in arrhythmogenesis and sudden cardiac death. Free running fibers are present inside the ventricle and connect to the myocardium at different entry points before subdividing to form a tree inside the cardiac muscle and allow propagation of the electrical impulse that triggers the mechanical contraction of the cardiac chambers. 3D imaging of its architecture¹ is therefore of high importance to better understand propagation patterns in cardiac arrhythmias and improve their diagnosis and treatment. Most commonly used techniques to study cardiac electrical conduction system remain destructive: histological sectioning, endocardial staining or ink injection². In order to address this challenging problem, we investigate the possibility to exploit high field (9.4T) MRI to acquire high resolution images of ventricular myocardium. Since Purkinje fibers (PF) have a distinct cellular structure and are surrounded by connective tissue, we propose to use the Magnetization Transfer (MT) imaging technique³. The aim of this study was to optimize MT parameters to enhance contrast between collagen fibers (surrounding PF) and myocardiumA 40 kg mal pig was anaesthetized and the heart was removed (sternal thoracotomy), flushed with cold cardioplegic solution and immersed in 4% formaldehyde in PBS. A sample of anterior left ventricle including interventricular septum was selected (3x4.5x8 cm) displaying several free-running PF in the endocardium at visual inspection. All experiments were performed at 9.4T/30cm (Bruker Biospin MRI, Ettlingen Germany). A cylindrical (72 mm inner diameter) 8 channels volume array Tx/Rx was used for ex vivo imaging. The cardiac structure was assessed through 2D T1w images with: TE/TR/matrix/FA NA/TA= 3.7/2000ms/225x150/75°/15/1 min15s; Resolution= 200x200 µm and a slice thickness of 0.5mm. A preparation module of MT was used and different parameters were tested: frequency offset Δf=2000, 3000, 5000 and 10000 Hz, RF field B1=5, 10 and 15 μT, finally the module duration D varied between: 50 to 450 ms. However, due to the maximal acceptable energy (0.5 W for the coil in continuous wave), MT module duration of 250, 350 and 450 ms at 15 μT could not be performed with this TR. Region of Interest were manually drawn and duplicated for each image: one surrounding the free PF in the cavity, one located in the parenchyma and one in the noise, respectively. Signal to noise ratio of PF and tissue were computed as the mean value of the signal in both ROI divided by the standard deviation in the noisy region. The MT contrast (MTC) between fiber and tissue was calculated as follow: MTC=$$$\frac{2(SNR_{f}-SNR_{t})}{SNR_{f}+SNR_{t}}$$$ Where SNR_f and SNR_t are the SNR of the fibers and tissue (parenchyma), respectively. With such a formulation, MTC is the contrast to noise ratio (CNR) divided by the mean SNR of both ROIs .Without MT module (Fig.1.1), almost no contrast appears on the image, whereas several tissue structures appear in presence of MT (Fig.1.2). MTC values were plot in function of D, B1 and Δf (graphs on fig.2). Whatever the MT pulse parameters, MTC was systematically found higher than without (0.17). However, MTC curves show different values depending on MT parameters, with a maximal value of 0.4 for an offset frequency around 3 to 5 kHz at a B1 amplitude of 10 µT and a pulse duration of 350 ms (red circle). Similar values were observed at this B1 value and Δf for pulse duration ranging between 250 and 450 ms, or for stronger B1 (15 µT) at a 150 ms pulse duration. Under these conditions, the increase of contrast compared to reference image is around 55%. Fig. 3 displays three orthogonal views extracted from the 3D acquisition performed with MT (Δf f=3 kHz, B1=10 µT and D=250 ms) at an isotropic resolution of 200 μm on the same sample. Insertion points from the free running PF can be visualized and their connexions followed inside the muscle (red arrows).First results of MT technique show an improved contrast in ex vivo heart tissue. Although all tissues exhibited a reduction in SNR in presence of MT, the contrast between PF and the cardiac tissue has been increased by more than 50% as compared to standard proton density weighted sequence. The decrease in SNR was expected since structural proteins (typically collagen) are present everywhere in the cardiac muscle. Fibrous structures are well defined in the myocardium and are associated to collagen fibers which are structural proteins surrounding PF.MT significantly improves contrast in ventricular myocardium and appears promising in imaging the 3D architecture of the Purkinje network.