Frequency-domain band-pass filtering enhanced pulsation method for pulmonary embolism diagnosis and risk stratification: a two-center retrospective study

Background Pulmonary embolism (PE) remains a major cause of morbidity and mortality in critical care, yet traditional diagnostic methods face limitations, especially in critically ill patients. This study introduces a novel frequency-domain band-pass filtering enhanced electrical impedance tomography (EIT) pulsation method for rapid, non-invasive PE diagnosis and risk stratification. Methods In a two-center retrospective study, 106 participants (53 PE patients, 53 healthy controls) were enrolled. A 16-electrode EIT system recorded pulmonary blood flow pulsation signals, and a heart rate-adaptive band-pass filter with its lower cutoff frequency set slightly below the estimated heart rate was applied to extract perfusion-related pulsatility while separating it from ventilation-related low-frequency components. Key parameters (Matching Index (MI), Dead space index (DI), Shunt Index (SI), Electrical Impedance VQ ratio(EIVQ)) were analyzed for PE diagnosis and risk stratification. Results Compared with healthy controls, PE patients had significantly lower MI ( P < 0.001) and higher DI ( P < 0.001) and SI ( P = 0.005). For risk stratification, intermediate-high-risk PE patients showed lower MI and higher DI than lower-risk patients, with the strongest differences versus low-risk patients (both adjusted P < 0.001).Among the EIT-derived parameters, the combined MI + DI+SI model achieved the highest AUC (0.820, 95% CI: 0.741–0.899), although the improvement over MI alone was marginal; it outperformed DI and SI. EIVQ showed no significant discriminative value. Conclusions The enhanced EIT pulsatility method shows promise for non-invasive bedside assessment of PE-related perfusion abnormalities. MI, DI, and SI differentiated PE patients from healthy controls, while MI and DI were associated with risk stratification. These findings support the diagnostic and risk-stratification potential of EIT-derived pulsatility parameters and warrant further validation in consecutive patients with suspected PE.