Just as magnetic resonance imaging has revolutionized clinical medicine, magnetic resonance microscopy has led to significant advances in the basic biomedical sciences, especially as research questions continue to arise. For example, “Scientists probing the connection between genotype and phenotype – the physical manifestation of the gene – need new tools to unwind the complex interactions,” said G. Allan Johnson, director of the Duke Center for In Vivo Microscopy. Here, confocal microscopy has proved a powerful imaging method. Light can penetrate to depths of up to 200 μm, which is sufficient to look at embryos of prenatal day nine or earlier. However, many complex changes take place in the critical days from prenatal days 10 to 15.“MR histology can provide unique three-dimensional insight,” he continued. “But, as with any new method, there is a steep learning curve to interpreting the images.” For this reason, in the Aug. 26 issue of PNAS, Johnson and colleagues at Duke University Medical Center, at the University of Edinburgh in the UK and at the National Institute of Environmental Health Sciences in Research Triangle Park, N.C., reported a four-dimensional atlas and morphological database they developed for the embryonic and neonatal mouse. “The online 4-D atlas provides the scientific community with the resources to climb that learning curve quickly,” he said.Five stages, three planesThe atlas identifies more than 200 structures in the mouse, in five stages in three planes. The researchers acquired the images from embryos from 10 prenatal stages – days 10.5 to 19.5, at one-day increments – and from mouse neonates at postnatal days 0, 2, 4, 8, 16 and 32. They scanned specimens up to postnatal day 8 at 9.4 T using an Excite console from GE Healthcare of Waukesha, Wis.; for neonates at postnatal days 16 and 32, they used a 7-T system with larger gradients accommodating the animals.The images in the atlas exhibit much higher spatial resolution than previous MRI atlases of the mouse embryo, which were acquired with much shorter scan times, Johnson said. “Previous work – including our own – has required very long scan times, with limited spatial resolution and limited field of view. One had to make the decision to cover only a small portion of the embryo at higher spatial resolution or cover the whole specimen at lower resolution. And whichever choice was made, the scans would require 10 to 15 hours. The methods were not readily scalable to larger numbers of specimens required to keep up with demand.”The researchers improved the spatial resolution by using active stains to reduce the spin lattice relaxation time, yielding a 10× signal enhancement. As is often the case, however, the solution created another challenge. Increases in spatial resolution are made possible by decreasing the size of the encoding voxel, which, in turn, increases the number of voxels required to cover a specimen, leading to much longer scan times. They addressed the challenge by using a novel encoding method to increase the amplifier gain in the high-frequency regions of the (Fourier) encoding space and enhance the details, coupling it with partial Fourier encoding to simultaneously increase resolution and decrease scan time to only three hours.Hard to shareThis increased the 3-D digital arrays to 1024 × 512 × 2048 pixels, 64 times larger than before, but led to still more challenges. “Once one attains such large arrays, it becomes very difficult to share them,” Johnson said. “There’s just too much data.” The researchers tackled this with VoxPort, a Structured Query Language database developed by MRPath Inc. (now Umlaut Inc.) of Research Triangle Park, N.C., that allows the virtualization of data. Thus, Web-based users receive only the data they need at the time they need it.Researchers have reported a 4-D mouse atlas that offers high-resolution images of more than 200 anatomical structures from 10 prenatal stages and six postnatal stages. Shown are representative images demonstrating how the atlas can be used. A volume-rendered image of a 14-day-old embryo shows the surface of the animal from the sagittal (a) or the coronal (b) perspective. The volume can be interactively sliced from any direction (c), revealing the complex anatomy of any structure — in this case, the heart. The 4-D atlas provides a normative data set showing the rendered normal heart with all four chambers (d). Note the normal septum between the left and right ventricles. Images acquired in an identical fashion of an (Smo -) knockout mouse demonstrate the septal defect (arrow) that is common with this genetic defect (e).The database can be accessed for free at http://www.civm.duhs.duke.edu/devatlas/index.html. Users can access the entire 4-D mouse atlas by registering and downloading software adapted from the Mouse Biomedical Informatics Research Network (MBIRN) Atlasing Toolkit. The atlas includes sagittal, transverse and coronal images of mouse embryos and neonates. The resolution is the same in all three cardinal planes, so users can define a plane along any arbitrarily chosen axis. The 200+ structures in the atlas are labeled in the three planes, at intervals of 195 to 585 μm, using accepted nomenclature for prenatal days 14.5 to 18.5. The annotated data is displayed using Atlasing Toolkit software.Simultaneous displays of images from multiple stages of development acquired with various imaging modalities is possible using VoxStation, a Java application. In addition, investigators can partition regions of interest, reconstruct anatomical structures using third-party software such as ImageJ and quantitatively assess changes in morphology.Johnson and colleagues demonstrated the utility of the atlas by showing how it can aid in the study of developmental changes in the heart. The PNAS paper includes coronal and transverse views of the heart from prenatal days 12.5 and 18.5 and from postnatal days 0 and 4, which were displayed simultaneously in the software, allowing users to match anatomical landmarks. Because larger structures, including the atria and ventricles, can be seen in the postnatal specimens, users can localize them in the earlier developmental time points. In addition, the isotropic resolution and multiple-plane views enable them to identify small structures such as the aortic, tricuspid and mitral valves as they develop.The researchers are working to expand the database of embryos to include a wide range of knockouts and mutants. “Data is being assembled in a common format and on our freely accessible database so scientists across the world can gain access to this technology,” Johnson said. The work was supported by the National Institutes of Health, National Center for Research Resources.