Evaluating Feature Extraction in Ovarian Cancer Cell Line Co-Cultures Using Deep Neural Networks

📁 Project Overview

Welcome to the project repository! This project consists of several components to handle image processing, feature extraction, and visualization from 2D co-culture assays. Here’s a brief overview of the folders and their contents:

1. CellProfiler Feature Extraction and Illumination Correction 🧪

This folder contains two main pipelines:

• Illumination Correction: Corrects for uneven illumination in raw images to ensure accurate feature extraction.
• CellProfiler Feature Extraction: Extracts single-cell features from cancer cells using CellPose masks. The features are then aggregated by well and exported as Excel files.

2. Data Preprocessing and Feature Extraction 🔄

Scripts in this folder are designed to ensure precise data extraction and correction, improving the quality of downstream analyses.

• Data Preprocessing Scripts: Includes merging fields of view, CellPose segmentation, and bounding box creation to prepare raw data for analysis.
• Feature Extraction: Python scripts for extracting single-cell features using the ResNet50 model.
• Enrichment Score Calculation: Code for calculating enrichment scores to assess the significance of features.

These scripts are used for preparing and processing data before the feature extraction and analysis stages.

3. R Visualization 📊

This folder contains R scripts for visualizing extracted features:

• UMAP and PCA: Dimensionality reduction techniques for visualizing complex data.
• Box Plots and Density Plots: Statistical plots to analyze and compare feature distributions.

These visualizations help in understanding the patterns and relationships within the data, providing insights into the results of your analyses.

---

Feel free to explore each folder to dive deeper into the components of this project.🚀