Abstract
Background
The aim of this study was to validate whether regional ventilation and perfusion data measured by electrical impedance tomography (EIT) with saline bolus could discriminate three broad acute respiratory failure (ARF) etiologies.
Methods
Perfusion image was generated from EIT-based impedance–time curves caused by 10 ml 10% NaCl injection during a respiratory hold. Ventilation image was captured before the breath holding period under regular mechanical ventilation. DeadSpace%, Shunt% and VQMatch% were calculated based on lung perfusion and ventilation images. Ventilation and perfusion maps were divided into four cross-quadrants (lower left and right, upper left and right). Regional distribution defects of each quadrant were scored as 0 (distribution% ≥ 15%), 1 (15% > distribution% ≥ 10%) and 2 (distribution% < 10%). Data percentile distributions in the control group and clinical simplicity were taken into consideration when defining the scores. Overall defect scores (DefectV, DefectQ and DefectV+Q) were the sum of four cross-quadrants of the corresponding images.
Results
A total of 108 ICU patients were prospectively included: 93 with ARF and 15 without as a control. PaO2/FiO2 was significantly correlated with VQMatch% (r = 0.324, P = 0.001). Three broad etiologies of ARF were identified based on clinical judgment: pulmonary embolism-related disease (PED, n = 14); diffuse lung involvement disease (DLD, n = 21) and focal lung involvement disease (FLD, n = 58). The PED group had a significantly higher DeadSpace% [40(24)% vs. 14(15)%, PED group vs. the rest of the subjects; median(interquartile range); P < 0.0001] and DefectQ score than the other groups [1(1) vs. 0(1), PED vs. the rest; P < 0.0001]. The DLD group had a significantly lower DefectV+Q score than the PED and FLD groups [0(1) vs. 2.5(2) vs. 3(3), DLD vs. PED vs. FLD; P < 0.0001]. The FLD group had a significantly higher DefectV score than the other groups [2(2) vs. 0(1), FLD vs. the rest; P < 0.0001]. The area under the receiver operating characteristic (AUC) for using DeadSpace% to identify PED was 0.894 in all ARF patients. The AUC for using the DefectV+Q score to identify DLD was 0.893. The AUC for using the DefectV score to identify FLD was 0.832.
Conclusions
Our study showed that it was feasible to characterize three broad etiologies of ARF with EIT-based regional ventilation and perfusion. Further study is required to validate clinical applicability of this method.
Trial registration clinicaltrials, NCT04081142. Registered 9 September 2019—retrospectively registered, https://clinicaltrials.gov/show/NCT04081142.
Electrical Impedance Tomography for Cardio-Pulmonary Monitoring
