PULSE 7B

🚀 PULSE-7B

Dataset for the paper "Teach Multimodal LLMs to Comprehend Electrocardiographic Images", aiming to advance ECG image interpretation.

🌐 Project Page: https://aimedlab.github.io/PULSE/ 📄 Paper: https://arxiv.org/abs/2410.19008 🧑‍💻 Code: https://github.com/AIMedLab/PULSE 👩‍⚕️ ECGInstruct(Training): https://huggingface.co/datasets/PULSE-ECG/ECGInstruct ⚖️ ECGBench(Testing): https://huggingface.co/datasets/PULSE-ECG/ECGBench

✨ Features

• Multimodal Design: PULSE-7B is a multimodal large language model (MLLM) dedicated to ECG image interpretation.
• Rich Dataset Utilization: It leverages the extensive ECGInstruct dataset with over one million instruction - tuning samples, enabling it to handle various ECG - related tasks from diverse data sources.
• Overcoming Traditional Limitations: Unlike traditional methods relying on raw physiological signals and limited to specific cardiac conditions, PULSE-7B can interpret both printed and digital ECG images, which is especially useful in resource - limited settings.
• Benchmark - Proven Performance: With the introduction of ECGBench, PULSE-7B sets new state - of - the - art performance, outperforming general MLLMs by an average accuracy improvement of 15% to 30%.

📚 Documentation

Introduction

We introduce PULSE-7B, a multimodal large language model (MLLM) specifically designed for ECG image interpretation. Leveraging the comprehensive ECGInstruct dataset, which contains over one million instruction - tuning samples, PULSE-7B is tailored to handle a wide range of ECG - related tasks drawn from diverse data sources. While traditional ECG interpretation methods are often constrained by their reliance on raw physiological signals and limited to specific cardiac conditions, PULSE-7B addresses these limitations by enabling robust interpretation of both printed and digital ECG images, making it especially valuable in resource - limited settings where access to raw signals may be restricted. In conjunction with the introduction of ECGBench, a benchmark that includes four key tasks spanning nine datasets, our experiments demonstrate that PULSE-7B establishes new state - of - the - art performance, surpassing general MLLMs with an average accuracy improvement of 15% to 30%. This model showcases the potential to significantly advance ECG image interpretation, providing a more versatile and accurate tool for clinical practice.

Overall Performance on ECGBench

The overall performance of PULSE-7B on ECGBench: 

Model Performance

In - domain


Out - of - domain


Case Study

Citation

If you find this work helpful, please cite our paper:

@article{liu2024teach,
  title={Teach Multimodal LLMs to Comprehend Electrocardiographic Images},
  author={Ruoqi Liu, Yuelin Bai, Xiang Yue, Ping Zhang},
  journal={arXiv preprint arXiv:2410.19008},
  year={2024}
}

📄 License

This project is licensed under the Apache - 2.0 license.