c-Fos Detection Code

This repository contains code for detecting c-Fos protein expression in images using a deep learning approach. The implementation employs a U-Net model to process multi-channel fluorescence images and segment c-Fos-positive cells.

Overview

The main images contain different colors and channels:

The code utilizes three channels from the main images:

• Alexa Fluor 488
• Alexa Fluor 594

• DAPI

The workflow includes the following steps:

1. Image Preprocessing:

   • High-resolution images are divided into smaller tiles of size 100x100 pixels for efficient processing.

     

2. Model Training:

   • The U-Net model is trained on these smaller image tiles, using the corresponding ground truth annotations for accurate segmentation.

3. Testing:

   • The trained model is tested using smaller chunks of images to evaluate performance.

     

4. Image Assembly:

   • After segmentation, the smaller tiles are assembled back into a larger image for comprehensive analysis.

   

Requirements

• Python 3.10.12
• torch
• cv2
• torchvision