Session: IMECE Undergraduate Research and Design Exposition

Paper Number: 121066

121066 - Image-Based Quantification and Identification of Live-Dead Cells Following Impact 

The long-term objective is to develop AI-based automation for neuron cell analysis and
identification of cell death after force in the perpendicular direction. Neurons, when active and
alive, connect to each other with dendrites. On a 2D gray-scale image taken with an
optical/fluorescence microscope, the neurons appear as long, connecting rods. Neurons, when
dead, appear as condensed circles, much smaller than their alive structure. Developing a routine
set of steps to identify the number of dead neuron cells efficiently and accurately in a live-dead
cell grayscale after impact image is the goal of this study. Currently, software, such as ImageJ,
can be used to count live-dead cells. The first step taken in the study was performing a
shear-force impact upon a small culture of the cell line, SH-SY5Y, cloned three times from a
subline of neuroblastoma cell line SK-N-SH. Multiple images and recordings were taken during
and after the impact. A gray-scale image was selected to use with ImageJ. Since ImageJ is
primarily used for image data analysis, specifically general measurements and analysis, the
number of dead neurons found in the gray-scale image needed to be counted manually
beforehand as a means for comparison. A total of 12 dead neurons were counted from the
gray-scale image. This study provides a more accurate means for identifying dead neuron cells
using ImageJ. The initial settings on ImageJ included changing the “Type” to “8-Bit,” changing
the “Process” to “Binary,” adjusting the “Threshold,” and changing the measurement range of
the area of a dead neuron cell (mm^2) to the smallest (.06 mm^2) and largest (.11 mm^2) dead cell
found. The “Binary” setting caused the gray-scale image to lose clarity and produce inaccurate
results. This setting caused Image J to count the total number of cells identified in the edited
image, regardless of the set range, 149 dead neurons. The next set of settings produced much
closer results. These steps included changing the “Type” to “8-Bit,” subtracting a total of 30
pixels from the background of the image, keeping the same measurement range for the area of a
dead neuron cell (.06 mm^2 - .11 mm^2), and adjusting the “Threshold.” The result is provided
on a data table generated by ImageJ with each dead neuron cell counted within the set
measurement range. While effective in accuracy to an extent, the process of counting these dead
neurons depends on the commands sent to ImageJ through a multitude of steps, making neuron
cell data collection from large-scale experiments inefficient and tedious. Further study can be
conducted to create set routines for identifying dead neurons from live-dead cell images of
different types in addition to phase contrast images. These routines can then be developed into a
program for AI to analyze without the need for manual counting and reference to increase
efficiency with data collection, such as the death rate of neurons, for large-scale experiments.

Presenting Author: Akanksha Subbarao Coppell High School, Summer Researcher at University of Texas at Arlington

Presenting Author Biography: Currently a Senior at Coppell High School, Coppell TX. She is a fellow of the New York Association of Biomedical Innovators / GynoEase Incorporation / Chief Administrative Officer since 2023. She has won the 2023 SWENext Global Innovator Award, auto admit into SWENext High School Leadership Academy, Student Member of the Biomedical Engineering Society (2023 - present), AP Distinction Award (2023), AP Scholar Award (2022), VIBHA Youth Bronze Star Award (2022): volunteer appreciation for 100+ service hours, Certified in OSHA 10-Hour General Industry Course (2020).

Authors:

Ashfaq Adnan University of Texas at Arlington
Akanksha Subbarao Coppell High School, Summer Researcher at University of Texas at Arlington
Raisa Akhtaruzzaman University of Texas at Arlington