Expediting image analysis for novel biological applications with an automated MATLAB algorithm

William Armstrong (JMU, Engineering '20) worked with me for 2 years to use image analysis tools in MATLAB to count number of cell nuclei present in a fluorescent microscope image. We worked together to find different applications for this, including working with Dr. Marta Bechtel (JMU, Biology) for a year to not only count how many cells were present, but also quantify the amount of actin present in her microscope images. Her work with cornea tissue showed that higher levels of actin (stained red in the images) meant the regenerating tissue was no longer transparent. Will's MATLAB code counted cells present and then recorded the intensity of actin present by looking at the red channel. He continued to work on his program by looking into multiple image analysis methods for counting cells and performed statistical ANOVA analysis to identify the "best" approaches based on parameters for the initial images. He looked at variables such as compactness and size through generating artificial high contrast images with known counts, and then compared the current image analysis quantification methods to the known counts.

Modeling the interaction of f-actin filaments

Melissa Riddle (JMU, Mathematics '20) worked on two projects over a few years with me. The first was looking at modifying the MATLAB Monte Carlo simulation from my postdoctoral work at UNC Chapel Hill to determine the mechanics and mathematical rules for how f-actin might bump or run into each other. The current simulation assumes that f-actin filaments exist in their own space, so can move through or past each other. Melissa used the principle of the coefficient of restitution from dynamics to determine how f-actin that bump into each other might then influence their ultimate movements.

Using MATLAB to quantify extracellular matrix fiber alignment by migrating cells

Christine Gatto (JMU, Engineering, '19) worked with images of ECM from Dr. Kristopher Kubow (JMU, Biology) that were organized by cells over time. The biology question centers around the complex feedback between ECM fibers and cells. Cells align ECM fibers, but which "signal" is the strongest for the cells? The previous alignment of the ECM or the cellular rearrangement of the ECM. Christine's project involved quantifying the ECM fibers. Before we could create an AI algorithm to identify the fibers, we needed to understand what we were trying to quantify, how the computer might quantify it, and how to then use the data to make sense of the results. Christine used MATLAB to identify fibers manually, and then worked with how to process the data of a series of points to determine a degree of alignment.