[공지사항]을 빙자한 안부와 근황 
Show more

Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

MEETI: A Multimodal ECG Dataset from MIMIC-IV-ECG with Signals, Images, Features and Interpretations

Created by
  • Haebom

Author

Deyun Zhang, Xiang Lan, Shijia Geng, Qinghao Zhao, Sumei Fan, Mengling Feng, Shenda Hong

Outline

MEETI (MIMIC-IV-Ext ECG-Text-Image) is the first large-scale multimodal ECG dataset that synchronizes electrocardiogram (ECG) waveform data, high-resolution ECG images, and detailed interpretation text generated by a large-scale language model. Each MEETI record consists of four components: raw ECG waveform, corresponding plot image, extracted feature parameters, and detailed interpretation text, which are aligned using a consistent unique identifier. This integrated structure supports transformer-based multimodal learning and enables fine-grained and interpretable inference on cardiac health. By bridging the gap between traditional signal analysis, image-based interpretation, and language-based understanding, MEETI lays a solid foundation for explainable next-generation multimodal cardiovascular AI and provides the research community with a comprehensive benchmark for the development and evaluation of ECG-based AI systems.

Takeaways, Limitations

•
Takeaways:
◦
Providing the first large-scale multimodal ECG dataset, laying the foundation for the development of multimodal AI systems
◦
Integrating raw signal, image, and text interpretation enables development of models that understand and integrate diverse ECG information in real-world environments.
◦
Support for transformer-based multi-modal learning and detailed and interpretable heart health inference
◦
Providing a comprehensive benchmark for developing and evaluating ECG-based AI systems
◦
Presenting a new research direction for electrocardiogram interpretation
•
Limitations:
◦
Lack of specific information about the size and diversity of the dataset.
◦
Additional validation of the quality and accuracy of the dataset is needed.
◦
Need to review reliability and bias of interpreted texts relying on large-scale language models
◦
Lack of comparative analysis with other similar datasets
👍
Made with Slashpage