Non-apnea electrical impedance tomography for bedside ventilation–perfusion assessment during spontaneous breathing: a prospective cross-over validation study

Conventional saline-contrast electrical impedance tomography (EIT) perfusion imaging typically requires a brief end-expiratory apnea during saline injection to reduce respiratory interference. However, this maneuver is often difficult or impractical in spontaneously breathing patients. We aimed to evaluate whether a non-apnea saline bolus EIT protocol could produce perfusion maps and ventilation–perfusion-related parameters similar to those obtained with the apnea-based reference protocol. Twenty adult patients who completed both study procedures were included: nasal cannula ( n = 7), high-flow nasal cannula (HFNC, n = 9), and invasive mechanical ventilation ( n = 4). Each patient underwent two saline-contrast EIT perfusion acquisitions: a reference end-expiratory apnea-based acquisition and a non-apnea acquisition during ongoing tidal breathing. The primary outcome was within-patient spatial concordance between perfusion maps, quantified by the pixel-wise Spearman correlation coefficient across lung pixels. Secondary outcomes included Bland–Altman agreement of global ventilation–perfusion-related parameters—specifically global inhomogeneity (GI) indices and percentages of EIT-derived V/Q categories (perfused, ventilated, and V/Q-matched units)—as well as regional correlations across four ventral-to-dorsal regions of interest (ROIs). Pixel-wise analysis demonstrated close within-patient spatial concordance in perfusion mapping ( r = 0.92; IQR, 0.88–0.93). Nineteen of 20 patients (95%) had an individual correlation coefficient > 0.80. Breathing-pattern variability did not differ statistically between protocols, as reflected by the standard deviation of tidal impedance amplitude ( P = 0.12) and respiratory cycle duration ( P = 0.68). Bland–Altman analysis showed small mean biases for GI, and percentages of EIT-derived V/Q categories, although limits of agreement were wider for some derived parameters. Regional analysis demonstrated high correlations for ventilation and perfusion across ROIs, with the strongest V/Q-match correlation in the dorsal dependent region (ROI 4; r = 0.960, P < 0.001). Repeated measures correlation confirmed strong overall within-subject regional associations ( r rm ranging from 0.703 to 0.915, P < 0.001). In this clinically heterogeneous cohort, the non-apnea protocol demonstrated strong within-patient spatial correlation with the reference apnea-based protocol for perfusion mapping; however, weaker agreement and wider limits of agreement for derived classified-pixel percentages support feasibility under the studied conditions without establishing clinical equivalence, diagnostic accuracy, or interchangeability.