Functional Segmentation for Electrical Impedance Tomography May Bias the Estimated Center of Ventilation

Functional segmentation of the region of interest (ROI) on electrical impedance tomography (EIT) images is performed by exclusion of voxels with volume variation below a specified threshold. Considering the heterogeneous spatial distribution of poorly aerated regions, this work assessed the ROI threshold influence on the ventilation distribution of healthy lungs. Sixteen adults were mechanically ventilated and positive end-expiratory pressure (PEEP) titrated (20 to 4 cmH₂O with 2 cmH₂O steps, 100 s each) while ventilatory and EIT data were recorded. ROI were delimited for each PEEP step based on the respective step (ROI_STEP) or the maximum PEEP (ROI_{MAX PEEP}), and on thresholds of 5%, 10%, 15% and 20%. Spatial distribution of ventilation was assessed by the center of ventilation (CoV) and the difference between higher thresholds (10%, 15% and 20%) and their respective 5% threshold counterpart (dCoV). Results showed positive correlation between CoV and PEEP for all strategies. While dCoV was not significantly different from zero for ROI_{MAX PEEP}, ROI_STEP dCoV was inversely proportional to PEEP and directly proportional to the chosen threshold. These results suggest that assessments of spatial distribution of ventilation might be biased by functional ROI delimitation parameters, especially for lower PEEP values.