Summer Internship Project | Indian Institute of Technology (IIT) Delhi
This repository presents the implementation and evaluation of deep learning models for Chest X-ray disease detection, completed during my Summer Internship at the Indian Institute of Technology (IIT) Delhi.
The project focuses on medical image classification using Chest X-ray images. During the internship, I studied the project pipeline, configured the development environment, resolved compatibility issues, trained and evaluated the models, and documented the complete workflow.
Note: This work is based on an existing open-source repository. My contribution includes environment setup, debugging, TensorFlow compatibility fixes, model training, evaluation, and documentation. I do not claim ownership of the original implementation.
- 🎓 Summer Internship at IIT Delhi
- 🧠 Deep Learning for Medical Image Analysis
- 🩻 Binary Chest X-ray Classification
- 🏥 Multi-label CheXpert Disease Classification
- ⚙️ TensorFlow/Keras Implementation
- 📈 Final Binary Classification Test Accuracy: 88.62%
- Understand deep learning techniques for medical image analysis.
- Implement and evaluate Chest X-ray disease detection models.
- Configure datasets and the project environment.
- Resolve TensorFlow compatibility issues.
- Train and evaluate deep learning models.
- Document the complete implementation and evaluation process.
HealthAI-Project/
│
├── backend/
├── dashboard/
├── models/
├── notebooks/
│ ├── 01_Binary_Chest_Xray_Classification.ipynb
│ └── 02_CheXpert_Multilabel_Classification.ipynb
│
├── results/
├── screenshots/
├── requirements.txt
└── README.md
- Dataset: Chest X-ray Pneumonia Dataset
- Task: Binary Image Classification
- Classes: Normal, Pneumonia
- Framework: TensorFlow / Keras
- Image Size: 224 × 224
- Batch Size: 32
- Optimizer: Adam
- Learning Rate: 1e-3
- Epochs: 20
| Metric | Value |
|---|---|
| Test Accuracy | 88.62% |
| Test Loss | 0.2623 |
- Dataset: CheXpert (Small)
- Task: Multi-label Chest X-ray Disease Classification
- Framework: TensorFlow / Keras
- Evaluation: Official validation split
Note: The public CheXpert (Small) dataset does not provide labeled test data. Therefore, evaluation is performed using the official validation dataset.
- Python
- TensorFlow
- Keras
- NumPy
- Pandas
- OpenCV
- Matplotlib
- Scikit-learn
- Jupyter Notebook
- Git
- Linux (Ubuntu)
- ✅ Binary Chest X-ray Classification – Completed
- ✅ Multi-label CheXpert Evaluation – Completed
- ✅ Repository Documentation – Completed
- ✅ Summer Internship Project – Completed
Clone the repository:
git clone https://github.com/Chandan9574/HealthAI-Project.gitMove into the project directory:
cd HealthAI-ProjectInstall the required dependencies:
pip install -r requirements.txtThe datasets are not included in this repository due to their large size.
Download the required datasets separately:
- Chest X-ray Pneumonia Dataset
- CheXpert (Small)
After downloading, place them inside the datasets/ directory before running the notebooks.
During this internship, I gained practical experience in:
- Deep Learning for Medical Imaging
- CNN-based Image Classification
- TensorFlow and Keras
- Model Training and Evaluation
- Medical Image Preprocessing
- Linux Development Environment
- Git and GitHub
- Debugging Machine Learning Projects
- Implementation and Evaluation of Deep Learning Models
This work was completed as part of my Summer Internship at the Indian Institute of Technology (IIT) Delhi.
I sincerely thank my mentor Darakshan Rashid and Prof. Brejesh Lall for their valuable guidance and continuous support throughout the internship.
The implementation is based on an existing open-source project, and all credit for the original repository belongs to its original author.
This repository is intended for educational and research purposes.
Please refer to the LICENSE file for additional information.
Chandan Kumar
B.Tech, Computer Science and Engineering
National Institute of Technology Mizoram
Summer Internship, IIT Delhi
GitHub Profile: https://github.com/Chandan9574