Segmentation of white matter ultrastructures in 3D electron microscopy

Abstract

We prepared ten samples from rats' white matter for serial block-face scanning electron microscopy (SBEM) imaging. We simultaneously acquired SBEM images of the white matter at low- and high-resolutions. The low-resolution datasets were acquired from big tissue volumes of 200 um x 100 um x 65 um with a voxel size of 50 nm x 50 nm x 50 nm. Two-thirds of the low-resolution images correspond to the corpus callosum and one-third to the cingulum. The high-resolution datasets were acquired from small tissue volumes of 15 um x 15 um x 15 um and imaged with a voxel size of 15 nm x 15 nm x 50 nm from the corpus callosum. All the images were acquired from the ipsi- and contralateral hemispheres of sham-operated rats (n = 2) and rats with traumatic brain injury (n = 3).

The high-resolution datasets were automatically segmented using ACSON pipeline [1] and the low-resolution datasets using DeepACSON pipeline [2].

Our segmentation of white matter ultrastructures enables quantifying axonal morphology, such as axonal diameter, eccentricity, and tortuosity, and the spatial organization of ultrastructures, such as the spatial distribution of myelinated axons and cell nuclei.

Please, find the "README" file for further explaining details of data.

You can find the "Methods" section in [1] and [2] for details related to SBEM image acquisition and image segmentation pipelines. In case of questions, please contact the corresponding author, and we will be happy to assist you.

Corresponding author: Alejandra Sierra

Please, cite the following articles if you use the datasets:

[1] Abdollahzadeh, A., Belevich, I., Jokitalo, E., Tohka, J. & Sierra, A. Automated 3D Axonal Morphometry of White Matter. Sci. Reports 9, 6084 (2019). DOI 10.1038/s41598-019-42648-2.

[2] Abdollahzadeh, A., Belevich, I., Jokitalo, E., Sierra, A. & Tohka, J. DeepACSON: Automated Segmentation of White Matter in 3D Electron Microscopy, Communications Biology, 2021