README.md

---
license: apache-2.0
language:
- en
library_name: transformers
pipeline_tag: feature-extraction
tags:
- medical
- cardiovascular
- ecg-image
- ecg-text representation learning
- ecg-foundation-model
- pytorch
---


<div align="center" style="font-size: 1.5em;">
  <strong>Learning ECG Image Representations via Dual Physiological-Aware Alignments</strong>
</div>

<div align="center" style="font-size: 2em;">
</div>

<div align="center">
<a href="https://arxiv.org/pdf/2604.01526" style="display:inline-block;">
    <img src="https://img.shields.io/badge/arxiv-Paper-red?style=for-the-badge">
  </a>
  <a href="/" style="display:inline-block;">
    <img src="https://img.shields.io/badge/Code-Github-blue?style=for-the-badge">
  </a>
  
  <a href="https://huggingface.co/Manhph2211/ECG-Scan" style="display:inline-block;">
    <img src="https://img.shields.io/badge/Checkpoint-%F0%9F%A4%97%20Hugging%20Face-White?style=for-the-badge">
  </a>
</div>




## Quickstart

```python
from transformers import AutoModel, CLIPImageProcessor
from PIL import Image
import torch

model = AutoModel.from_pretrained("Manhph2211/ECG-Scan", trust_remote_code=True)
model.eval()

processor = CLIPImageProcessor.from_pretrained("openai/clip-vit-large-patch14-336")
img = Image.open("ecg.png").convert("RGB")
pixel_values = processor(images=img, return_tensors="pt")["pixel_values"]

with torch.no_grad():
    out = model(pixel_values).embeddings         
```

## Citation

```bibtex
@article{pham2026learning,
  title={Learning ECG Image Representations via Dual Physiological-Aware Alignments},
  author={Pham, Hung Manh and Tang, Jialu and Saeed, Aaqib and Ma, Dong and Zhu, Bin and Zhou, Pan},
  journal={arXiv preprint arXiv:2604.01526},
  year={2026}
}
```