Blood Cell Classification Project

Table of Contents

1. Introduction
2. Datasets
   • BCCD Dataset
   • Complete Blood Count (CBC) Dataset
   • PBC_dataset_Normal_DIB
3. Project Structure
4. Setup and Installation
5. Usage
   • Data Preparation
   • Model Training
   • Model Evaluation
6. Acknowledgements
7. License
8. Contact

Introduction

This project focuses on the classification of blood cells using a convolutional neural network (CNN) model implemented with TensorFlow. The datasets used in this project are BCCD Dataset, Complete Blood Count (CBC) Dataset, and PBC_dataset_Normal_DIB,

Datasets

BCCD Dataset

BCCD Dataset is a small-scale dataset for blood cells detection. The dataset includes three types of lables: RBC (Red Blood Cell), WBC (White Blood Cell), and Platelets.

Overview

• Image type: JPEG
• Image size: 640x480
• Annotations: VOC format XML files

For more details, see For more details, see the BCCD Dataset README.

Complete Blood Count (CBC) Dataset

The CBC dataset contains 360 blood smear images along with their annotation files, split into training, testing, and validation sets. It includes images annotated for RBC, WBC, and Platelets.

Overview

• Image type: JPEG
• Image size: 640x480
• Annotations: VOC format XML files

For more details, see the CBC Dataset README.

PBC_dataset_Normal_DIB

This dataset contains 17,092 images of individual normal cells, organized into eight groups:

• Eosinophils
• Lymphocytes
• Monocytes
• Neutrophils
• Basophils
• Platelets
• Immature granulocytes
• Erythroblasts

The images were acquired using the analyzer CellaVision DM96 in the Core Laboratory at the Hospital Clinic of Barcelona and annotated by expert clinical pathologists.

Overview

• Image type: JPEG
• Image size: 360x363
• Annotations: in file names

For more details, see the PBC_dataset_Normal_DIB on Mendeley Data.

Project Structure

.
├── BCCD_dataset
│   ├── BCCD
│   │    ├── Annotations
│   │    ├── ImageSets
│   │    └── JPEGImages
│   ├── dataset
│   │   └── mxnet
│   │       ├── prepro.py
│   │       └── test.py
│   ├── scripts
│   │   ├── split.py
│   │   └── visualize.py
│   ├── example.jpg
│   ├── export.py
│   ├── LICENCE
│   ├── plot.py
│   ├── README.md
│   └── test.csv
├── Complete-Blood-Cell-Count-Dataset
│   ├── Testing
│   │   ├── Annotations
│   │   └── Images
│   ├── Training
│   │   ├── Annotations
│   │   └── Images
│   ├── Validation
│   │   ├── Annotations
│   │   └── Images
│   ├── _config.yml
│   ├── LICENSE
│   └── README.md
├── Model
│   ├── my_model
│   │   └── assets
│   ├── variables
│   │   ├── variables.data-00000-of-00001
│   │   └── variables.index
│   ├── fingerprint.pb
│   ├── keras_metadata.pb
│   └── saved_model.pb
├── PBC_dataset_Normal_DIB
│   ├── basophil
│   ├── eosinophil
│   ├── erythroblast
│   ├── ig
│   ├── lymphocyte
│   ├── monocyte
│   ├── neutrophil
│   └── platelet
├── unified_dataset
│   └── Images
├── .gitignore
├── LICENSE
├── Blood_Cell_Classification.ipynb
├── requirements.txt
└── README.md

Setup and Installation

1. Clone the repository:

git clone https://github.com/yourusername/your-repo-name.git
cd your-repo-name

2. Create and activate a virtual environment:

python3.8 -m venv venv
source venv/bin/activate

3. Install the required packages:

pip install -r requirements.txt

Usage

Data Preparation

Data preparation scripts are provided in the Blood_Cell_Classification.ipynb notebook.

Model Training

The model training script is provided in the Blood_Cell_Classification.ipynb notebook.

Model Evaluation

The model evaluation script is provided in the Blood_Cell_Classification.ipynb notebook.

Acknowledgements

• BCCD_Dataset: Thanks to Shenggan and Nicolas Chen for providing the BCCD Dataset.
• Complete Blood Count (CBC) Dataset: Thanks to MahmudulAlam for providing the CBC Dataset.
• PBC_dataset_Normal_DIB: Thanks to the authors of the PBC_dataset_Normal_DIB dataset.

License

This project is licensed under the MIT License - see the LICENSE file for details. Note: The datasets used in this project have their own licenses. Please refer to the original sources for more information.

Contact

If you have any questions or suggestions, please feel free to contact me at bv2340@columbia.edu.