DeepCardio

Background & Challenge

Atrial fibrillation (AF) is the most common cardiac arrhythmia and can lead to serious complications, such as stroke, systemic embolism, and heart failure. Paroxysmal atrial fibrillation (PAF) is one of the critical AF types that should be detected and managed early.

PAF is diagnosable when a 10-second, 12-lead ECG shows an irregular rhythm. However, early detection faces significant hurdles:

• Low detection yield due to the intermittent and often asymptomatic nature of the condition.

• Difficulty during Normal Sinus Rhythm (NSR): Detecting PAF on ECGs during NSR is especially challenging because the RR interval and P-wave are often similar to those of healthy individuals, making accurate diagnosis even harder.

Our Solution: The TS-ECG Network

In this study, we introduce a deep learning model designed to assist cardiologists in the early diagnosis of PAF during NSR by classifying ECGs into Healthy-NSR ECG and PAF-NSR ECG categories.

We propose the total-sub-length ECG (TS-ECG) network, a deep learning model that simultaneously learns two distinct features manifesting in AF:

• Total-length framework: Focuses on the rhythm characteristics across the entire length of the ECG.

• Sub-length framework: Focuses on learning the subtle P-wave segments.

Data Collection & Verification

• Dataset: We collected 3,591 ECGs from 785 PAF patients and 4,385 ECGs from 1,583 healthy individuals.

• Verification: We verified the performance of the proposed model on ECG data from healthy individuals with no history of AF and from those who experienced PAF during NSR at a multicenter university hospital in South Korea.

Results: Model Performance

On the PAF-NSR test dataset from Inha University Hospital, the model achieved outstanding diagnostic accuracy.

GitHub Repository: https://github.com/mskim1024/TS-ECG