🧬 White Blood Cell Classification & Morphological Feature Extraction

Assignment 1 – Introduction to Deep Convolutional Neural Networks
Goal: White Blood Cell (WBC) Classification + Morphology Detection

---

📌 Project Overview

This project implements a Deep Convolutional Neural Network (CNN) capable of performing two tasks:

1️⃣ Classifying White Blood Cells (WBCs)

The model predicts the WBC type (e.g., neutrophil, eosinophil, lymphocyte, monocyte).

2️⃣ Extracting Key Morphological Attributes

The system also predicts clinically relevant features such as:

• Cell Shape
• Nucleus Shape
• Cytoplasm Vacuoles

This project uses TensorFlow 2 with Keras and follows a multi-output CNN architecture.
Training and evaluation are done in a Jupyter Notebook environment, optimized to run both locally and on AWS.

---

🚀 Features

• Multi-task Deep CNN (classification + morphology)
• End-to-end training pipeline
• Image preprocessing using OpenCV & TensorFlow
• Data augmentation for better generalization
• Evaluation using accuracy, F1-score, confusion matrix
• TensorFlow 2.12 (Keras) workflow
• AWS-friendly environment setup

---

🛠️ Tech Stack

• TensorFlow 2.12
• Python 3.9
• OpenCV
• NumPy & Pandas
• Matplotlib & Seaborn
• Scikit-learn
• JupyterLab

---

📁 Project Structure

WBC-MultiTask-CNN/
│
├── data/
│   ├── raw/               # Original images
│   ├── processed/         # Preprocessed images
│   └── labels.csv         # Label file
│
├── notebooks/
│   └── assignment_1.ipynb # Main training & evaluation notebook
│
├── models/
│   └── wbc_cnn.h5         # Saved trained model
│
├── environment.yml        # Conda environment file
├── requirements.txt       # Package dependencies
├── README.md              # Project documentation
└── .gitignore

---

🔧 Installation & Setup

1️⃣ Clone this repository

git clone https://github.com/sheikhmunim/wbc-multi-task-cnn.git
cd wbc-multi-task-cnn

2️⃣ Create the Conda environment

conda env create -f environment.yml
conda activate wbc-cnn-env

3️⃣ Launch JupyterLab

jupyter lab

---

🧪 Training the Model

Open the notebook:

notebooks/main.ipynb

Inside the notebook, follow the guided steps:

• Load dataset
• Preprocess images
• Build the CNN architecture
• Configure multi-output losses
• Train the model
• Evaluate the results

---

📊 Evaluation

Model performance is measured using:

• WBC Classification Accuracy
• Morphological Feature Accuracy
• F1-score
• Confusion Matrix
• Training & validation curves

These metrics assess generalization to unseen microscopic images.

---

☁ Deployment (AWS Ready)

This setup is compatible with:

• AWS EC2
• AWS Sagemaker Notebook Instances
• GPU-enabled EC2 instances

---

📌 Future Improvements

• Expand dataset for more morphology labels
• Experiment with deeper CNN designs
• Use multi-branch CNNs for morphology prediction
• Add deployment interface (FastAPI / Streamlit)

---

📜 License

This project is for academic use under RMIT University (Assignment 1).

---