---
license: mit
library_name: pytorch
tags:
- Medical Vsion-Language Pre-Training
- BenchX
---

# PTUnifier Checkpoint Model Card

A retrained PTUnifier model for benchmarking medical vision-language pre-training methods within the BenchX framework.

## Model Details
- **Model Type**: PTUnifier
- **Architecture**: CLIP-ViT-Base image encoder and RoBERTa-Base text encoder
- **Original Papers**: [Towards Unifying Medical Vision-and-Language Pre-training via Soft Prompts](https://arxiv.org/abs/2302.08958)
- **Benchmark Paper**: [BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays](https://arxiv.org/abs/2410.21969)
- **Benchmark Framework**: https://github.com/yangzhou12/BenchX

## Intended Use
- **Primary Use Cases**:
  - Benchmarking performance for Medical Image Classification
  - Benchmarking performance for Medical Image Segmentation
  - Benchmarking performance for Medical Report Generation

## Pre-Training Data
- **Dataset**: 
  - Data source(s): MIMIC-CXR
  - Types of medical images: Frontal chest X-rays
  - Text data type: Associated radiology reports

## Prerequisites

Please follow the [instruction](https://github.com/yangzhou12/BenchX/blob/release/README.md#installation) to install BenchX. 

## Training & Evaluation

### 1. Classification

Fine-tuning PTUnifier for classification is not supported yet.

### 2. Segmentation
To fine-tune PTUnifier for segmentation, run this command:

```
python mmsegmentation/tools/train.py config/benchmark/<dataset_name>/ptunifier.yml
```

### 3. Report Generation
To fine-tune PTUnifier for report generation, run this command:
```
python bin/train.py config/report_generation/<dataset_name>/ptunifier.yml
```

### 4. Evaluation
To evaluate fine-tuned PTUnifier models, run:

```
# For classification and report generation
python bin/test.py config/<task_name>/<dataset_name>/PTUnifier.yml validator.splits=[test] ckpt_dir=<path_to_checkpoint>

# For segmentation
python mmsegmentation/tools/my_test.py mmsegmentation/config/<dataset_name>/ptunifier.yml <path_to_checkpoint>
```

## Citations
```bibtex
@inproceedings{chen2023towards,
  title={Towards Unifying Medical Vision-and-Language Pre-training via Soft Prompts},
  author={Chen, Zhihong and Diao, Shizhe and Wang, Benyou and Li, Guanbin and Wan, Xiang},
  booktitle={Proceedings of ICCV},
  pages={23403--23413},
  year={2023},
}
```
```bibtex
@inproceedings{zhou2024benchx,
  title={BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays},
  author={Yang Zhou, Tan Li Hui Faith, Yanyu Xu, Sicong Leng, Xinxing Xu, Yong Liu, Rick Siow Mong Goh},
  booktitle={Proceedings of NeurIPS},
  year={2024}
}
```