Myocardial infarction (MI) is associated with apoptosis of cardiomyocytes (CM) and differentiation of cardiac fibroblasts (CF) into myofibroblasts (MF). The latter leads to an increased production of extracellular matrix (ECM) proteins and to scar formation [1]. In order to characterize cardiac tissues and their remodelling, we introduced new methods for mounting of samples as well as for image acquisition and 3D analysis of tissue microstructure at submicrometer scale. Our new approach for tissue mounting has two advantages: (1) avoiding morphology changes induced by ample mounting and (2) an increased penetration depth during imaging. We obtained left ventricular tissue biopsies from an Institutional Animal Care and Use Committee [2] approved rabbit MI model. Animals were anesthetized prior to surgery by IM injection of ketamine (50mg/kg) and xylazine (10mg/kg). Narcosis was maintained by isoflurane. Infarction was induced by ligation of the circumflex artery as described in [3]. Additionally, fentanyl patches (25µg/h) were applied to all animals 24 h before until 48 h after the surgery. Animals were euthanized by IV injection of sodium pentobarbital (100mg/kg) before hearts were excised and tissue biopsies taken. After fixation and cryosectioning, we fluorescently labelled ECM, cell nuclei, α-smooth muscle actin (αSMA), and vimentin (Vim). Specificity of these labels for identification of CF and MF is limited [4]. Thus, αSMA antibodies label not only MF, but also cells in the wall of arterioles. Vim labels CF but also other cell types located adjacent to capillaries. We applied a new approach for embedding samples in a mounting medium without compression. In contrast to our prior approaches, probes were mounted on glass slides and sealed without additional cover glass slides. Scattering in the sample was largely removed by controlled desiccation in a humidity chamber before sealing. We acquired 3D images with a linear increase in laser power to compensate for depth-dependent attenuation. An algorithm for attenuation correction under conditions of linear power increase was developed and applied by fitting an extended attenuation model to the mean fluorescence intensity recorded in each plane. For all image channels, histogram based segmentation (mode + 2 standard deviations) was performed, followed by semi-automatic segmentation of CM and blood vessels using the water shedding algorithm [5]. Vessel lumens were dilated by ~1 µm to reconstruct vessel walls. Positively stained for Vim, these were excluded from segmented CF and MF. Our approach resulted in extended imaging depths and improved tissue morphology preservation, compared to conventional methods [5,6] (Fig. 1B versus A). The prior approach led to compression of samples, which made it difficult to identify vessels and flattened CMs. With the new approach capillary lumens and laminae of connective tissue were identifiable and more realistic CM shapes were maintained. We conclude that our new approach allows for improved quantitative characterization of cardiac tissue morphology.