Self-tuned object tracking algorithm for live-cell bright field time-lapse microscopy
Guardado en:
| Autor: | |
|---|---|
| Formato: | tesis de maestría |
| Fecha de Publicación: | 2022 |
| Descripción: | This thesis presents a novel self-tuned object tracking algorithm specifically designed for live-cell brightf ield time-lapse microscopy. The primary motivation for this project stems from the challenge of analyzing cell behavior over time using bright-field microscopy images, which are typically more challenging to process compared to fluorescence microscopy due to lower contrast and noise. The Damage Proliferation Phenomenon (DPP), a significant contributor to multiresistant and aggressive cancer phenotypes, plays a central role in this investigation. The proposed tracking algorithm leverages machine learning and pattern recognition techniques to address this challenge. By utilizing Bayesian optimization and gradient descent methods, the algorithm achieves robust performance in tracking cells across multiple time-series images, offering a solution to limitations posed by traditional manual observation and segmentation methods. This tool can significantly aid in understanding cancer cell behavior, improving diagnosis, and informing treatment strategies. The study includes a comprehensive review of the current state of the art, comparisons of various segmentation algorithms, and tests the developed algorithm using real-world datasets from live-cell microscopy. Results show improvements in tracking accuracy, particularly in detecting DPP-related behaviors, and provide insights into potential future enhancements. |
| País: | Kérwá |
| Institución: | Universidad de Costa Rica |
| Repositorio: | Kérwá |
| Lenguaje: | Inglés |
| OAI Identifier: | oai:kerwa.ucr.ac.cr:10669/100069 |
| Acceso en línea: | https://hdl.handle.net/10669/100069 |
| Palabra clave: | ALGORITHM MICROSCOPY ANALYSIS CELL BEHAVIOR |