🦴 ADLM: Addressing Class Imbalance in Bone Tumor X-Ray Classification with Generative Models

This project explores how generative models can help address class imbalance in bone tumor X-ray classification.
We use the BTXRD dataset, which contains X-ray images of different primary bone tumor entities.

---

🎯 Project Goals

• Implement a ResNet-based CNN as a baseline model for tumor classification.
• Apply and analyze class imbalance handling techniques such as Weighted Loss, Focal Loss, and Data Augmentation.
• Use generative models (e.g., GANs or Diffusion Models) to create synthetic X-ray images and evaluate their impact on classification performance.

---

🦴 Target Tumor Types

We focus on classifying seven tumor types:

• Osteochondroma
• Osteosarcoma
• Multiple Osteochondromas
• Simple Bone Cyst
• Giant Cell Tumor
• Synovial Osteochondroma
• Osteofibroma

---

🏫 About the Project

This project is part of the Advanced Deep Learning Methods (ADLM) course at the
Technical University of Munich (TUM), in collaboration with the
Clinic for Orthopaedics and Sports Orthopaedics and the
Institute for AI and Informatics in Medicine.

---

⬇️ Dataset Setup

Place the BTXRD dataset under the following paths relative to the project root:

data/
  dataset/
    BTXRD/
      images/         # Original X-ray images (e.g., IMG000123.jpeg)
      Annotations/    # JSON annotation files (same basenames as images)

Optional folders created by scripts in this repo:

data/
  dataset/
    final_patched_BTXRD/          # Extracted patches from annotations (created, this is the final dataset used for training and testing)
    squared_padded/         # Padded originals for 106 special cases (created)
    squared_patched_106/    # Patches from padded images (created)
    patched_BTXRD_merged/   # Merge of the two patch sets (created)

Install dependencies:

pip install -r requirements.txt

---

▶️ How To Run (Preparation → Training → Testing)

1. Extract patches from annotations

python data/btxrd_bounding_box_dataset_extractor.py

This creates data/dataset/final_patched_BTXRD/ from BTXRD/images + BTXRD/Annotations.

2. Train (with optional early stopping)

python src/training_ResNet.py --model resnet50 --early-stop --patience 10 --min-delta 0.001

Notes:

• The training pipeline reads labels directly from JSON annotations and splits in-memory.
• By default it uses data/dataset/patched_BTXRD_merged/ if present, otherwise falls back to patched_BTXRD/.
• Checkpoints are written under checkpoints/<model>/.

3. Test and generate confusion matrix

python src/testing_ResNet.py --model resnet50

Outputs:

• checkpoints/<model>/test_predictions.npy
• checkpoints/<model>/confusion_matrix.png

4. (Optional) Quick visualization of predictions

python src/plot_predictions.py

---

**▶️ How to Run SupCon Loss **

1. Contrastive Pretraining (run train_supcon.py)

2. Linear Classifier Training (run train_linear.py)

3. Evaluation (run eval_supcon.py)

Outputs:

• checkpoints_supcon/<time>/encoder_supcon.pth
• checkpoints_linear/<time>/classifier.pth

ℹ️ Notes

• CSVs like dataset_singlelabel.csv are not required for training/testing in this pipeline; labels are taken from annotation JSONs. If needed for analysis, you can generate a CSV aligned to the patched images via:

python data/create_csv_patched.py