Study of Echocardiogram Parameters from PPG Signal Using Self-Organized Operational Map-based Network

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Echocardiography parameters like LVEF and MPI are vital for diagnosing heart conditions.
• Current diagnostic methods often require specialized equipment and settings.

Purpose of the Study:

• To explore the prediction of LVEF and MPI values from Photoplethysmography (PPG) signals.
• To evaluate patient classification based on LVEF and MPI.
• To investigate the potential of PPG for non-invasive cardiac health assessment.

Main Methods:

• Feature extraction from PPG signals.
• Utilized a Dual Attention-Self Organised Operational Map-LSTM-Conv Network (DASLCN) for analysis.
• Employed Self-Organized Operational Maps (SOOM) for feature mapping, followed by BiLSTM and 1D CNN layers.

Main Results:

• Achieved a regression value of 0.86 with 5.32±8.9% error for MPI prediction.
• Obtained an accuracy of 0.90 and sensitivity of 0.89 for LVEF prediction.
• Demonstrated successful regression and classification within the same network.

Conclusions:

• PPG signals can potentially predict key echocardiography parameters (LVEF, MPI).
• The DASLCN model shows promise for accurate, non-invasive cardiac diagnosis.
• This approach offers a cost-effective and portable alternative to traditional echocardiography.