EIT Outperforms Quantitative CT in Stratifying ARDS Severity After Lung Transplantation: A Retrospective Study

Respiratory mechanics and gas exchange parameters are readily accessible indicators for evaluating ventilation in patients with acute respiratory distress syndrome (ARDS). Computed tomography (CT) is a widely used radiological imaging technique, while electrical impedance tomography (EIT) is a novel technique for pulmonary function monitoring developed in recent years. Studies suggest that EIT combined with quantitative CT holds unique value in assessing ventilation/perfusion (V/Q) function. Unlike classic ARDS, the pathophysiologic alterations in ARDS following lung transplantation are complex, and the specific mechanisms remain unclear. This study aims to explore the use of respiratory mechanics, gas exchange parameters, EIT, and quantitative CT for the multimodal assessment of V/Q function in post-lung transplantation ARDS. We further aim to integrate visualizing dynamic imaging and regional quantitative lung analysis into this multimodal assessment system.

We retrospectively enrolled lung transplant recipients admitted to the intensive care unit who met the Berlin criteria for ARDS. Inclusion required an arterial partial pressure of oxygen to fraction of inspired oxygen ratio (P/F) ≤ 300 mmHg, with both EIT monitoring of V/Q and high-resolution CT performed within 24 h of documented P/F ≤ 300 mmHg. Patient baseline characteristics, respiratory mechanics parameters, gas exchange parameters, EIT data, and CT images were collected. Subjects were stratified into two groups according to P/F values: a low P/F group (P/F < 200 mmHg) and a high P/F group (200 mmHg ≤ P/F ≤ 300 mmHg). Ventilation parameters derived from EIT included global inhomogeneity index (GI), center of ventilation (COV), and regional ventilation delay index (RVDI). Using hypertonic saline contrast-enhanced EIT, we acquired V/Q parameters and calculated both global and regional EIT-based dead space fraction (EIT-Dead Space), intrapulmonary shunt fraction (EIT-Shunt), and V/Q matching (EIT-V/Q Match). Chest CT images were processed through a multitask learning U-net-based computer-aided diagnostic model. This enabled semiautomated lung segmentation, identification of high-density lesions, quantitative analysis, and three-dimensional visualization. Pulmonary volumes, lesion volumes, and percentage lesion volumes (calculated as lung lesion volume divided by lung volume) were, respectively, quantified for the left and right lungs.

The study ultimately enrolled 21 lung transplant recipients with ARDS, comprising five patients in the low P/F group and 16 in the high P/F group. Among 21 patients, the low P/F group demonstrated significantly higher ventilation ratio (VR), RVDI, and EIT-Dead Space, along with lower EIT-V/Q match levels compared to the high P/F group, while no significant difference in EIT-Shunt was observed between the two groups. EIT-Dead Space showed substantial agreement with ventilator-measured dead space fraction, and VR exhibited a significant positive correlation with EIT-Dead Space. In the two groups, quantitative CT-derived pulmonary metrics-including lung volume, lesion volume, and percentage lesion volume-showed no significant differences between the low and high P/F groups.

Among lung transplant recipients with ARDS, the low P/F group demonstrated elevated VR, RVDI, and EIT-Dead Space, alongside reduced EIT-V/Q matching levels when compared to the high P/F group. Notably, no significant differences were found in the quantitative CT-derived lesion volume parameters between the two groups.