Nuclei Segmentation Project

This repository contains code for nuclei segmentation using two different approaches: Detectron2 and YOLO v8. It also includes data handling scripts for preprocessing and converting data between different formats.

Repository Structure

.
├── 336a_detectron2_nuclei_segmentation.ipynb
├── 336b_yolov8_nuclei_segmentation.ipynb
├── data_handling_code/
│   ├── 01-remove_unwanted_data.py
│   ├── 02-Convert_label_masks_to_COCO JSON.py
│   ├── 03-Convert_COCO JSON_to_YOLOv8.py
│   ├── 04a-visualize-COCO labels.py
│   ├── 04b_visualize-COCO labels_filled.py
│   ├── 04c-visualize-YOLO labels.py
|   └── 04d-visualize-YOLO labels-filled.py
|    
└── README.md

Segmentation Models

336a: Detectron2 Nuclei Segmentation

This script (336a_Detectron2_Instance_Nuclei.ipynb) implements nuclei segmentation using Facebook's Detectron2 framework. Key features include:

• Installation of Detectron2 and required dependencies
• Dataset registration using COCO format
• Model configuration and training
• Evaluation using COCO metrics
• (Optional) Inference on test images and saving results

336b: YOLO v8 Nuclei Segmentation

The script 336b_training_YOLO_V8_Nuclei.ipynb implements nuclei segmentation using the YOLO v8 model. (Note: Details about this script are not provided in the given content.)

Data Handling Code

The data_handling_code folder contains scripts for various data preprocessing and conversion tasks:

1. 01-remove_unwanted_data.py: Scripts for initial data preparation and cleaning from Kaggle.
2. 02-Convert_label_masks_to_COCO JSON.py: Tools to convert label masks to COCO JSON format.
3. 03-Convert_COCO JSON_to_YOLOv8.py: Scripts to convert COCO JSON annotations to YOLO format.
4. 04(a-d)-Visualize: Tools for visualizing the labeled data in different formats.

Acknowledgments

• All credits go to Bhattiprolu S. who has worked on creating a wonderful Youtube channel explaing how to get kickstarted with machine vision and image segmentation for biologists.

References

1. Mahbod, A., Polak, C., Feldmann, K., Khan, R., Gelles, K., Dorffner, G., Woitek, R., Hatamikia, S., & Ellinger, I. (2023). NuInsSeg: A fully annotated dataset for nuclei instance segmentation in H&E-stained histological images. arXiv preprint arXiv:2308.01760. https://arxiv.org/abs/2308.01760
2. Bhattiprolu, S. (2023). python_for_microscopists. GitHub. https://github.com/bnsreenu/python_for_microscopists/tree/master/336-Nuclei-Instance-Detectron2.0_YOLOv8_code