🏥 Automatic Medical Report Generation

This project explores deep learning methods for generating medical reports from chest X-ray images using the Indiana University Chest X-ray dataset.

🧠 Architectures Implemented

1. ResNet50 (Encoder) + GPT-2 (Decoder) — PyTorch
2. DenseNet121 (Encoder) + LSTM (Decoder) — TensorFlow

Both models are trained, evaluated, and include report samples.

---

📁 Project Structure

notebooks/

├── 01_resnet50_gpt2.ipynb # PyTorch: ResNet50 + GPT-2 (with output & evaluation)

├── 02_densenet121_lstm.ipynb # TensorFlow: DenseNet121 + LSTM (with output & evaluation)

---

📊 Model Evaluation

Each notebook includes:

• BLEU Scores
• ROUGE Metrics
• Generated medical reports
• Visual and qualitative comparison of both architectures

---

🛠️ How to Run

1. Clone the repository

   git clone https://github.com/yaekobB/medical-report-gen.git
   cd medical-report-gen
   

2. Install dependencies

   pip install torch tensorflow transformers matplotlib

3. Open the notebooks Use Jupyter Notebook, Colab, or VSCode with Python support.

📦 Dataset

Dataset Name: Chest X-rays (Indiana University)

Kaggle Link: https://www.kaggle.com/datasets/raddar/chest-xrays-indiana-university

📌 Note: This dataset is not included in the repo. Download it from Kaggle and adjust the path in the notebook if needed.

📷 Example Output

Included in notebooks:

Sample generated reports

Evaluation scores

Output visualizations

📄 License This project is open-sourced under the MIT License.

📬 Feel free to reach out or open an issue with feedback or contributions!

---

Let me know if you'd like me to turn this into a README with badges, links to Colab, or sample image previews!