Multi-Objective Undercomplete Independent Component Analysis for Radar Signals based Heart Rate and Interbeat...

Area of Science:

• Biomedical Signal Processing
• Independent Component Analysis
• Cardiovascular Monitoring

Background:

• Extracting cardiac signals from mixed sources is challenging.
• Traditional methods may struggle with accuracy and robustness.
• Undercomplete signal separation offers a potential solution.

Purpose of the Study:

• To develop a Multi-Objective Undercomplete Independent Component Analysis (MO-UICA) method.
• To reconstruct cardiac signals from in-phase and quadrature components.
• To accurately estimate average heart rate and interbeat intervals.

Main Methods:

• Proposed MO-UICA method treats heart rate extraction as an undercomplete problem.
• Utilizes a multi-objective cost function combining cdf entropy and autocorrelation.
• Employs the Levenberg-Marquardt algorithm for optimization using gradient information.

Main Results:

• Demonstrated efficacy and robustness in estimating average heart rate.
• Successfully estimated interbeat intervals across various processing windows (1-60 seconds).
• Quantitative and qualitative results validate the proposed MO-UICA approach.

Conclusions:

• The MO-UICA method provides an effective approach for cardiac signal reconstruction.
• The technique shows promise for accurate heart rate and interbeat interval estimation.
• Future work will focus on enabling instantaneous heart rate estimation.