Introduction Histo-anatomy is a major determinant of heart function, affecting behaviours such as spread of excitation, stress/strain distribution, contraction/relaxation, and coronary perfusion, to name but a few. A detailed quantitative understanding of cardiac structure-function interrelation in norm and disease therefore requires an equally detailed insight into the histo-anatomical make-up of the whole organ (Burton et al. 2006). Thus far, partial histological reconstructions have been used in conjunction with diffusion tensor magnetic resonance (MR) imaging, to provide local high-resolution information for MR validation. However, these data sets are inherently limited by the resolution of the MR techniques used, usually yielding voxel dimensions for the whole heart in the range of 10^-3-10^-4 m. Here, we demonstrate an approach to mapping cardiac anatomy at 10^-5-10^-6 m resolution for an entire rabbit heart. Preparation A female New Zealand white rabbit was terminally anesthetised using 140 mg/kg intravenous sodium pentobarbital, in accordance with the Home Office Animals (Scientific Procedures) Act of 1986. The heart was excised, rinsed in heparinised (25 IU/L) Tyrode solution, and cardioplegically arrested using high potassium (20 mM KCl) modified Tyrode. The heart was fixed using freshly prepared Karnovsky’s fixative (a 2% paraformaldehyde / 2.5% glutaraldehyde mix). After high-resolution anatomical MR imaging, the heart was dehydrated using increasing concentrations of alcohol and wax embedded over a three-week period. Longitudinal sections of the entire heart were taken at 10 μm thickness using a heavy duty sledge type microtome (Leica SM2400). Sections were relaxed in a water bath (Leica HI 12110) at 40°C before being placed onto numbered 3-aminopropyltriethoxysilan coated slides. Each section was stained using the Masson’s Trichrome technique, before being mounted using DPX and left to dry in a fume hood overnight. Imaging All sections were imaged using a Leica QWin workstation and Leica QGO software to obtain mosaic images of the entire section (the largest sections were up to ~40 mm long and ~20 mm wide). Using a 10x objective, final in-plane pixel resolution was 1.1 µm x 1.1 µm. A total of 1,850 sections were produced (whole heart thickness was 18.5 mm), yielding voxel dimensions of 10^-5 x 10^-6 x 10^-6 m. Imaging time depended on tissue section size and varied from 30 s to over 10 min. Individual images were stored as bitmaps (BMP) ranging from a few hundred megabytes (MB) to just under 2 gigabytes (GB) in size. Challenges While developing the above method, a number of challenges were identified, including: – Difficulty in avoiding air bubbles in the complex cardiac structure during embedding (mild vacuum and changes in sample orientation needed). – Incomplete tissue relaxation (15-20 min per section can be required). – Terabytes (TB) of disk space are necessary to store the images. Compression and conversion to the PNG format of individual images was explored; the time required (1-3 min for compressing and 20-50 s for decompressing 1-2 GB images, respectively) made this unattractive for local use, though it may be necessary to facilitate sharing of information over the Internet (the compression results in an image file size reduction to 50-70% of the original). – Larger images cannot be rendered natively in Microsoft Windows. To this end, we developed a BMP viewer which can be downloaded from http://mef.physiol.ox.ac.uk/Software/BMPViewer.exe. The software creates a thumbnail that is automatically generated when selecting a BMP file. A second window is used to render only part of the image at its native resolution. Clicking anywhere on the thumbnail updates the contents of the native viewer, which can also be panned around the image. The total volume of histological information for one heart was just over 1.3 TB. This is currently being used, in conjunction with the MR data obtained prior to sectioning, for reconstruction of partial tissue areas, such as ventricular tissue wedges or papillary muscle (Plotkowiak et al., 2008). In the process, techniques for whole heart reconstruction are generated.