Image Analysis To Classify Cells

Students

2021
Engineering and Applied Sciences

Faculty

Patricia M. Williams Term Professor in Biology; Co-Director, Penn Genomics Institute; Adjunct Professor, Computer and Information Scienc

Project Summary

This summer, I learned and experienced various depths of image analysis techniques to investigate and understand single cell activities. In Biology, classification of single cells is important in disease prevention and cancer diagnostics as proportions of different types of cells can provide useful information about an illness. There are so many cells in a human body that can be distinguished from one another by simple phenotypic differences such as shape, size and weight. But, single cells have similar shapes and size which make it a tough task to classify with high accuracy and speed. That’s why we use image analysis to classify these cells.

My mentor was working on a particular type of cells called CD8+ T cells. When we are attacked by pathogens, there are certain types of cells called memory cells that remain in our body after the pathogen is removed so as to fight it stronger and faster when they reappear. It is important to fight pathogens faster as they can make us sick or disrupt some cell functions in our body which could lead to death. Many researches have proved that memory cells originate from effector T cells which are the cells used to fight pathogens. But, researchers haven’t figured out which effector T cells become memory cells and which ones die and leave the body.

After getting many images over time of effector T cells from my mentor, I had to do image analysis on them to see if there are any changes in cells over time so as my mentor can research on it more. I learnt about Image segmentation which is the process to separate cells in an image to get information of all the individual cells. I also learnt and applied image filtration and normalization. This was an important part of my project as many cells in the images had illumination variations and noise which made it hard to get accurate information. After getting information of individual cells such as mean, mode, Kurt, and many more I developed an algorithm in MATLAB to draw graphs of those information over time to see if there are any changes over time. Throughout the project, I was trying different filtrations and cell detection algorithms to get accurate results.

I understood research as very goal oriented before doing this project and I would expect to get things done before time. But, after not getting the predicted results over time I realized research is more about wandering around and trying to make something work. I gained an insight of how a team works especially when results aren’t obtained. I enjoyed my research a-lot and I hope to do it in my future.

.