🏥 Intraoperative Zonular Video Analysis

📂 Project Structure

The project is organized into Autonomous Agents and Vision-Language Models (VLMs) modules.

share/
├── agent/                        # Autonomous Agents
│   ├── aide/
│   │   ├── best_solution.py
│   │   └── run_aide_experiment.py
│   ├── codex/
│   │   └── train_surgical_classifier.py
│   ├── copilot/
│   │   └── train_model.py
│   └── qoder/
│       └── surgical_image_classifier.py
│
└── vlm/                          # Vision-Language Models
    ├── gemini.py
    ├── gpt.py
    ├── gemma.py
    ├── qwen.py      / qwen_seq.py
    ├── qoq_med.py   / qoq_seq.py
    ├── huatuo.py    / huatuo_seq.py
    ├── lingshu.py   / lingshu_seq.py
    ├── internvl.py  / intern_seq.py
    │
    ├── HealthGPT/                # Submodule: HealthGPT
    │   └── scripts/
    ├── LLaVA-Med/                # Submodule: LLaVA-Med
    │   ├── llava.py
    │   └── llava/
    └── OphthaReason/             # Submodule: OphthaReason
        └── OR.py

💾 Data Availability

⚠️ Restriction: The dataset used in this study is restricted to academic research purposes only.

You can access the dataset via Hugging Face:

• Dataset Link: Intraoperative-Zonular-Video-Dataset

🛠️ Environment Setup

General Requirements

To set up the base environment, please install the dependencies listed below.

pip install torch==2.8.0 torchvision==0.23.0
pip install transformers==4.57.1 huggingface-hub==0.34.4 rich==14.0.0

Model-Specific Dependencies

Some VLMs utilize their own codebases. Please refer to their respective repositories for detailed installation instructions:

✏️ Citation

If you find this code or our dataset useful for your research, please consider citing our study:

@article{ZonularBench, title={Vision-Language Models vs Autonomous AI Agents for Anterior Capsular Radial Folds: A Diagnostic Study}, url={https://www.medrxiv.org/content/early/2026/01/23/2026.01.15.26344200}, DOI={https://doi.org/10.64898/2026.01.15.26344200}, publisher = {Cold Spring Harbor Laboratory Press}, author={Zhang, Ye and Chen, lei and Zhao, Wenhao and Zhang, Hui and Qiao, Chunyan and Liu, Zijian and Chung, Charlotte Hui and Tan, Marcus Chun Jin and Wang, Meng and Tham, Yih Chung and Koh, Victor and Cheng, Chingyu and Liu, Dianbo}, year={2026}, month={Jan}, journal = {medRxiv} }

‌