#Serial images serial#
#Serial images registration#

In order to detect reliable correspondences, both block matching and SIFT require that the adjacent section images cannot change too much.

#Serial images serial#

Scale Invariant Feature Transform (SIFT) 11 is also frequently used to identify correspondences across serial sections 12. Each section is modeled as a triangular spring mesh grid, and block matching is used to search for the corresponding positions of the grid vertices between the adjacent sections.

#Serial images registration#

Elastic volume registration method 10 makes assumption that the shape of biological tissue changes smoothly across adjacent sections and the non-linear distortion suffered in every section is independent. Therefore, it makes sense to calculate the transformations of serial images simultaneously. As a result, these methods may introduce error accumulation and propagation, which leads to artificial deformation for the images far away from the reference image. Some methods choose one of serial section images as the reference image, and then do forward or backward pairwise image registration sequentially 9. Several non-linear registration methods have been proposed to align serial section electron microscopic images. As a result, image registration is a challenging task to stitch the section images into a 3D volume 8. Furthermore, sections suffer stretching, folding and distorting individually during section preparation and imaging. However, the continuity of the tissue structures between sections is destroyed when the tissue block is cut into sections physically with anisotropic resolution. These sections can be imaged in parallel to save the time of data acquisition which is desirable for large volume reconstruction. For serial sections imaging 6, 7, the sections are cut in an ultra-microtome with 30-50 nm thickness, collected onto conductive supporter, such as electron opaque support tape, and imaged in EM. However, the methods are destructive to the biological tissue since the sections are lost as soon as they are removed from the block face, and it may waste all the previous efforts with interruption. As the section images are acquired in situ, the stacks are aligned inherently. The block surface is imaged in EM, and then removed with high precision up to 5 nm using a diamond knife 4 or Focused Ion Beam (FIB) 5. For block-face EM imaging, tissue block is fixed in EM vacuum chamber. Each of the available methods involves tradeoffs in terms of resolution, acquisition speed, and reliability. In general, there are two different methods for volume EM imaging, which are block-face EM imaging and serial sections EM imaging.

Fortunately, with the development of electron microscopy (EM) in recent years, the ability of image acquisition is no longer the obstacle for volume reconstruction of 3D biological tissue at synaptic level 2, which can provide new in-depth insights for neuroscience 3. Optical microscopy cannot provide sufficient resolution to reveal these fine structures 1.

To image and analyze neuronal circuit at synaptic level, obtaining three-dimensional nervous tissue with high resolution always appeals to neuroscientists. We demonstrate that our method outperforms the state-of-the-art methods on serial EM sections including a synthetic test case. The proposed method is highly automatic and could recover the 3D continuity of the neurite. Finally, the sections are warped according to the adjusted positions of skeleton points. In company with the constraints that the displacements of the skeleton points in the same section are smooth and small, an energy function is proposed to calculate the new positions of the skeleton points for all of the sections. Then we make assumption that the skeleton shape is smooth adequately in z direction. The virtual skeletons are traced across the sections after an initial approximate rigid alignment. To get round the difficulty of reliable landmark extraction, we propose a skeleton-based image registration method for serial EM sections of the nerve tissue. The traditional methods use the SIFT or block matching method to extract landmarks between the adjacent sections, which is doubtful when the neurite direction is not perpendicular to the section plane. As a result, image registration is a challenging task to recover the continuity of the neurite. However, the continuity between sections is destroyed when the tissue block is cut into sections physically, and sections suffer stretching, folding and distorting individually during section preparation and imaging. Imaging serial sections in electron microcopy (EM) is an important volume EM approach for neuronal circuit reconstruction, which has advantages of larger imaging volume and non-destructive for tissue sections.