3D estimation of synaptic vesicle distributions in serial section transmission electron microscopy

Research output: Contribution to conferenceConference abstract for conferenceResearchpeer-review

To transfer information between neurons, synaptic vesicles move toward the presynaptic membrane, called the active zone, and fuse with it, releasing neurotransmitters into the synaptic cleft. Thus, the shortest distance from vesicles to the active zone affects the speed of signal transportation directly. It is hypothesized that in a rat model of behavioral stress the vesicles distribution varies. We propose methods for estimating the 3-dimensional distribution of synaptic vesicles from the active zone through serial section transmission electron microscope images (ssTEM) from Sprague-Dawley rat brains.

We demonstrate that ssTEM images have an additive bias field, which is well modelled by a quadratic polynomial. ssTEM images make a 3D study on very high-resolution images possible. However, due to the physical cutting of a section from a 3D embedded tissue, the relations between sections are lost. To reconstruct the 3D data we register the images in a common coordinate system. The traditional method to measure the distribution of the vesicles is to measure the distance independently of neighbouring sections. This is biased depending on the slope of the active zone with respect to the section. We suggest two alternatives to estimate: 1) the bias and correct for it in an existing estimated distribution; 2) the shortest distance from the 3D reconstruction.

The proposed method has been applied to five datasets of ssTEM images of male rat brains including 123 images. After intensity inhomogeneity correction and image registration, we obtained the histogram of minimum distances of vesicles to the active zone with the traditional method and two proposed improvements. Comparing the traditional method with either of the two suggested methods using Wilcoxon signed rank test, shows a strong difference in the results with p-values less than 10^(-10) in both cases.

We conclude that the two proposed modeling significantly improves the measures on the estimated synaptic vesicle distribution in relation to the active zone.
Original languageEnglish
Publication date8 Jun 2013
Number of pages1
Publication statusPublished - 8 Jun 2013
EventEuropean Congress of Stereology and Image Analysis - Kaiserslautern, Germany
Duration: 8 Jul 20139 Jul 2013
Conference number: 11


ConferenceEuropean Congress of Stereology and Image Analysis

ID: 85365119