Detection and Classification of Chromosomes with Sister Chromatid Cohesion Defects Using Object Detection Models

This repository will be the sample code for the title paper.

Repository Structure

1. Data: Metaphase Images

• images: Microscopic images of metaphase spreads from DDX11 -/- cells.
• labels: Bounding box coordinates and class labels (typeA : 0, typeB : 1, typeC : 2) provided in YOLO format.

2. Models: Trained Object Detection Models

This folder contains trained weight files (.pt) for YOLOv8n, used to detect and classify chromosomes based on their morphological cohesion patterns.

• Architecture: YOLOv8n
• Classes:
  • typeA: Normal cohesion
  • typeB: Partial cohesion defect (arm dissociation)
  • typeC: Complete cohesion defect (total separation)

3. Code: Detection_Classification.ipynb

A Google Colaboratory notebook designed for automated detection and classification.

Usage

1. Open the Notebook: Locate Detection_Classification.ipynb in the Code folder.
2. Launch in Colab: open in colab
3. Run Pipeline: Execute the cells in order from top to bottom.