Introducing the ARGO dataset of post-ischemic ventricular tachycardia bipolar electrograms

Orrù, Marco;Baldazzi, Giulia;Bandino, Stefano;Zirolia, Davide;Bertagnolli, Livio;Viola, Graziana;Pani, Danilo

2023-01-01

---

Abstract

The development of automatic tools to identify, characterize, and delineate abnormal ventricular potentials (AVPs) in post-ischemic ventricular tachycardia (VT) electrograms (EMGs) is still an open research issue. As a significant amount of annotated data is required for this purpose, the lack of open and annotated datasets hampers the research progress in this field. This work introduces a new open dataset that will be made available on PhysioNet by the end of 2023, which was recorded, annotated, and validated in the framework of the “Ablation Reinforcement by computer aided Guidance and Optimization: ARGO study” project. The proposed dataset comprises 2034 anonymized recordings from nine post-ischemic VT patients, including 2.5 s bipolar EGMs, unipolar EGMs, the 12-leads surface ECGs, and the electroanatomic maps (voltage and local activation time maps). The dataset annotation was carried out by three expert electrophysiologists, who categorized each bipolar EGM into the classes of Physiological, AVP, or Unknown/Undetermined, and delineated the onset and end of the AVP. The ARGO dataset represents a valuable tool to enable the effective development and benchmarking of AVP detection and delineation algorithms, their characterization, but also to encourage collaboration across the research community.