Cardiac and respiratory signal separation in electrical impedance tomography (EIT).

Electrical Impedance Tomography (EIT) is a promising non-invasive imaging modality for respiratory and cardiac monitoring. However, separating cardiac signals from the dominant respiratory component remains challenging because both physiological processes are simultaneously embedded in the measured impedance signals. This study investigates the influence of EIT measurement patterns on cardiac-respiratory signal separation using four methods: Continuous Wavelet Transform (CWT), Empirical Mode Decomposition (EMD), Independent Component Analysis (ICA), and Principal Component Analysis (PCA). Numerical simulations were performed in EIDORS using a realistic thoracic geometry reconstructed from the POPI dataset with a GREIT-based image reconstruction framework. Four measurement configurations were evaluated: adjacent-adjacent, adjacent-Skip 4, Skip 4-adjacent, and Skip 4-Skip 4. Signal separation performance was assessed using the Cross Correlation (CC) between the extracted and reference signals, while the normalized Root Mean Square Error (RMSE) was additionally used to evaluate the reconstructed cardiac waveform. Among all evaluated methods, PCA consistently achieved the highest waveform similarity, reaching a Lung CC of 98.12\% and a Heart CC of 99.56\% under the Skip 4-Skip 4 measurement pattern. Visual inspection of the reconstructed images further demonstrated clearer separation of respiratory and cardiac conductivity distributions using this configuration. These results indicate that combining PCA with a wide electrode measurement pattern improves cardiac-respiratory signal separation in simulated EIT measurements.