Automated detection of ultrastructural features at neuronal synapses

Synaptic vesicles are the ultracellular structures responsible for carrying chemical messengers known as neurotransmitters from inside the axon of a neuron to the synaptic junction outside. The variation in size and location of these structures is important in the study of their use and reuse in neurons. We propose a method to locate and estimate the diameter of vesicles in electron microscope images of synapses. We train a U-Net inspired model to perform pixel-wise segmentation of the vesicles against background pixels. We then use contour detection on the resulting segmentation maps to determine individual vesicle centers and effective diameters. To our knowledge, there are no baselines in this task so we establish one on an in-house dataset. Our results show that the proposed model performed well on this task.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2020

Author dc:creator

• Ramesh, Ashwin

Contributors dc:contributor

• Koyejo, Oluwasanmi

• Vesicle
• Detection
• Segmentation
• Synaptic
• Synapse
• Neuronal
• Ultrastructural
• U-Net
• Features

Statement dc:rights

• Copyright 2020 Ashwin Ramesh

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/108195

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/108195