Semi-Automated Image Quantification of Cell Proliferation in Gilt Uterus Tissue

This repository contains two MATLAB scripts developed for image-based segmentation and quantification of cell proliferation in gilt uterus tissue, using histological images stained with Ki67. This project was completed under the mentorship of Dr. Uduak George at San Diego State University, in collaboration with colleagues at Purdue University.

---

Project Overview

The purpose of this project was to automate the quantification of proliferating versus non-proliferating cells in gilt uterus tissue, collected during a study on postnatal colostrum intake. Two MATLAB algorithms were developed:

1. Image Segmentation Segments tissue into mucosa, connective, and muscle regions using hand-drawn ROIs.

2. Cell Area Quantification Applies color thresholding to Ki67-stained images to calculate the area of proliferating and non-proliferating cells.

This research was part of a broader effort to understand how early colostrum intake (10% vs 20% of body weight) affects postnatal tissue growth and potential fertility in gilts.

---

Scientific Background

• Sample source: Gilt uteri from Purdue University's swine research farm
• Staining: Ki67 for proliferating cells, H&E as visual reference
• Treatment groups: COL10 and COL20, based on % body weight in colostrum intake
• Analysis: Boxplots, t-tests, and random forest classification used to assess impact
• Findings: No statistically significant difference in proliferation between groups, but biological trends and other significant indicators (e.g., immunocrit, TEMP-24H) were revealed

---

Technologies Used

• MATLAB (R2020b - Version 9.9.0.1467703)
• Image Processing Toolbox (Version 11.2)
• Color Thresholder App (HSV-based masking)
• Adobe Photoshop (for panorama assembly)

---

How to Use

Requirements:

• MATLAB R2020b or later
• Image Processing Toolbox

---

Step 1: ROI-Based Tissue Segmentation

1. Navigate to the folder:
   1) Image Segmentation - ROI Dissection

2. Run the script inside to manually draw Regions of Interest (ROIs) that segment the tissue image into:

   • Mucosa
   • Connective
   • Muscle

3. Use the H&E-stained image as a guide for accurate anatomical ROI drawing.

Segmented images will be saved for use in the next step.

---

Step 2: Tissue Area Masking and Cell Quantification

1. Navigate to the folder:
   2) Tissue Area Masking

2. Run the following MATLAB scripts in this order:

   • areaMain.m – Initializes the analysis pipeline
   • colorAreaCalculator.m – Applies HSV-based color thresholding
   • proliferationMask.m – Masks proliferating (brown) cells
   • nonproliferationMask.m – Masks non-proliferating (blue) cells
   • totalTissueAreaMask.m – Calculates total tissue area

---

Output Files

• Segmented tissue images for:
  • Mucosa
  • Connective
  • Muscle
• Binary masks of:
  • Proliferating cells
  • Non-proliferating cells
• An Excel summary containing:
  • Proliferating cell area
  • Non-proliferating cell area
  • Total tissue area for each tissue type

---

Poster Presentation

This project was presented at the 2021 SDSU Student Research Symposium (SRS).

Click the poster image to view the full PDF version.

---

Attribution & Acknowledgments

• Images provided by:

  • Dr. Theresa Casey, Department of Animal Sciences, Purdue University
  • Dr. Ariany Suarez-Trujillo & Kelsey Teeple

• Mentor: Dr. Uduak George (SDSU)

• Authors:

  • Brooke Tyler
  • Sashiel Vagus (GitHub Profile)

• Funding: Supported by NSF PUMP Research Grant No. DMS-1916494

---

Citation

Tyler, B., & Vagus, S. et al. (2021). Predictive Multi-Scale Modeling of Postnatal Regulation of Protein Synthesis in Gilts. The PUMP Journal of Undergraduate Research.

---

Contact

Sashiel Vagus
San Diego State University
Email: svagus2@sdsu.edu
GitHub: @sashielvagus

---

License

This repository is shared under an academic research license. For permission to reuse or cite the work, please contact the author(s).