High-Resolution Episcopic Microscopy (HREM) is a technique for visualising tissue samples, such as mouse embryos, as a large stack of successive 2D images. The resolution of the images approaches that of conventional histology and, since they are perfectly aligned, they can be used to construct very detailed 3D models.
As with conventional histology, HREM imaging uses samples that have been embedded in a rigid matrix and sectioned with a microtome. However, the technique does not require sections to be individually stained or registered. Instead, each image is obtained from the exposed surface of the tissue during the sectioning process.
The embedding medium is a hard plastic resin, which enables sections as thin as 1 μm to be accurately removed. The resin itself contains fluorescent dyes and, under fluorescent illumination, tissue at the cut surface of the block can be easily visualised against the bright background of the plastic. By sequentially imaging the block face during the sectioning process (a technique known as episcopic imaging), a comprehensive stack of accurately aligned images is acquired, documenting the 3D structure of the sample. The only practical constraints on the 2D image resolution are the choice of optics used to visualise the block surface and the capability of the camera capturing the images.
Because HREM does not require the collection, staining and registration of individual sections, samples can be imaged rapidly. For example, the E14.5 mouse embryo data on this site is based on 3 μm sections, a process that takes around 8 hours and generates approximately 3,500 images for each embryo.
Building detailed 3D models from images of individual sections has always proved difficult, since accurate alignment of the images is required. In addition, the sections themselves often become distorted as they are cut, captured and stained.
Using HREM avoids these problems by sequentially imaging the face of the block, rather than imaging individual cut sections. As a result, the resolution of the 3D data is only constrained by the choice of section thickness, since this determines the frequency of images that will provide the 3D volume.
HREM images capture different tissues and structures as distinctive patterns and ‘textures’ of pixels, using a broad range of grayscale values. As a result, the data lends itself to simple, 3D volume rendering as is commonly used in medical imaging. It can also be used as the template for visualising individual structures traced through successive images, modelled by isosurface rendering.
HREM data can be analysed either as 2D image stacks or by 3D rendering. The 2D images are routinely used by the DMDD programme to identify detailed morphological phenotypes in developing mouse embryos. Thanks to the intrinsic alignment of the images, orthogonal and oblique views of the tissue sample can also be calculated from HREM data with relatively little loss of resolution. These alternative views provide valuable additional information about the morphology of a sample.
Since HREM sections can be cut as thin as 1 μm, 3D models of HREM images typically have voxel resolutions of 1-8 μm3. At this resolution it is possible to identify features such as individual nerves and blood vessels, which may not be detectable using lower-resolution techniques such as optical projection tomography (OPT), micro-magnetic resonance imaging (μMRI) or micro-computed tomography (μCT).
Next: Sample preparation
DMDD does not provide an imaging service, but we can help or advise you about HREM. Please email firstname.lastname@example.org.