Cell Segmentation in Mouse Brain Images
Date of Award
12-12-2024
Degree Name
M.S. in Computer Science
Department
Department of Computer Science
Advisor/Chair
Tam Nguyen
Abstract
The mouse cerebellar cortex, with its layered structure of molecular, Purkinje, and granular layers, refines motor control, coordination, and balance. Purkinje cells, the primary output neurons, process signals from parallel and climbing fibers and send inhibitory outputs to cerebellar nuclei, enabling motor learning and cognitive processing. Nowadays we have a better understanding of the cerebellar cortex's crucial roles in brain function and its implications in neurological disorders thanks to recent developments in neuroimaging, optogenetics, and electrophysiology that have clarified mechanisms underlying synaptic plasticity and signal processing within the region. Segmenting the cerebellar cortex is crucial for analyzing its cellular structure and connectivity, helping us understand motor control and learning processes. It enables detailed study of neural circuits and is key for investigating structural changes related to neurological disorders, aiding in targeted research and potential treatments. In this thesis, we apply various segmentation models to identify mouse brain cells through instance segmentation, focusing on precise classification of cell types and structures within the cerebellar cortex. We collect and annotate a dataset of 1000 images of mouse brain cells. We further evaluate different instance segmentation methods in two families, namely, transformer and non-transformer-based methods. Results indicate that non-transformer models outperformed transformer models in the small sized cells whereas transformer-based methods achieve better performance on large-sized cells. This thesis paves way to further research in using computer vision to understand mouse brain cells.
Keywords
Machine Learning, Segmentation, Instance Segmentation, Semantic Segmentation
Rights Statement
Copyright © 2024, author.
Recommended Citation
Shrivastava, Aditya, "Cell Segmentation in Mouse Brain Images" (2024). Graduate Theses and Dissertations. 7487.
https://ecommons.udayton.edu/graduate_theses/7487
