Detecting Apnoea in Premature Neonates: Exploring Frequency-Domain and Power Analysis of Electrical Impedance Tomography Lung Impedance Signals

Abstract Neonatal apnoea is often defined as a cessation of breathing, a critical challenge in premature infant care, demanding early and accurate detection. This study integrates Electrical Impedance Tomography (EIT) with frequency-domain analysis to identify apnoea-related changes in lung impedance signals. We processed lung impedance waveforms from fifteen premature infants (median gestational age: 31 weeks) and assessed frequency, amplitude, and power as potential biomarkers. Statistical analysis using two-way ANOVA (p < 0.05) demonstrated significant changes in impedance features before, during, and after apnoea episodes. We further evaluated each feature’s classification performance via Receiver Operating Characteristic (ROC) curves, reporting Area Under the ROC Curve (AUC), sensitivity, specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV). Frequency alone achieved a high AUC (0.98), with 85.96% sensitivity and 96.67% specificity, outperforming amplitude (AUC = 0.81) and power (AUC = 0.92). Notably, amplitude and power combined (AUC = 0.98) performed similarly to frequency on AUC but better on other measures (98.33% specificity). Similarly, combining power and frequency improved sensitivity, 89.47% at 97.50% specificity, surpassing frequency alone on both metrics. These results demonstrate the value of utilising a variety of attributes. While amplitude changes may be expected during apnoea, adding frequency and power metrics offers a more robust, multidimensional view of lung impedance behaviour. Moreover, residual signals during apnoea likely stem from cardiac activity and may also reflect slight changes in heart rate. This underscores the novelty of isolating lung-specific impedance from overlapping physiological signals. In conclusion, combining EIT and frequency-domain analysis offers a promising non-invasive method for real-time apnoea detection. In neonatal critical care units, it emphasises using multi-feature techniques to improve diagnostic accuracy.