Machine Learning and Simulation for High-Throughput Single-Cell Measurement
Author(s)
Malta, Nathan
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
The following work concerns increasing throughput and reducing manual labor required when conducting biological experiments. As a prerequisite for a wide range of experiments in biology, cells must be cultured. During culturing, important parameters regarding the health of these cells must be either approximated by eye or manually counted in microscope images. In the first chapter of this work, I introduce a novel approach to detecting cells in microscope images and extracting quantitative, biologically relevant parameters to aid researchers.
The remainder of this work considers patch clamping, an important technique for characterizing the electrical changes of individual, living cells. In this technique, a high-precision robotic manipulator is driven inside of a cell, so that electrical recordings can be taken. This technique is currently being used to better understand a wide array of ailments, including Alzheimer's disease and macular degeneration. Unfortunately, the technique has been plagued by low-throughput. To improve automation and throughput I create new high-accuracy calibration routines and an automatic patch clamping protocol.
Finally, I attempt to tackle the high barrier to entry for learning about patch clamping. A patch clamping rig cost many tens or hundreds of thousands of dollars due to the specialized equipment needed. This severely limits the number of people familiar with the technique. To address this, I create a software simulation, a video game, that walks a user through the highlights of the technique. This software is open source and free to download. Preliminary user trials suggest that users develop a better understanding of the technique in just minutes of playing.
Sponsor
Date
2023-12-05
Extent
Resource Type
Text
Resource Subtype
Thesis