Data Segmentation Using Principal Component Analysis of Pulmonary Perfusion Assessed by Electrical Impedance Tomography
摘要
In this study, we implemented techniques to analyze data obtained from electrical impedance tomography examinations to evaluate pulmonary perfusion using hypertonic saline solution injections. Our research utilized Principal Component Analysis to analyze 51 perfusions in 8 healthy pigs under mechanical ventilation, producing masks for assessing both hybrid and pulmonary regions of interest. Given the electrical impedance tomography's broad field of view covering heart and lung effects, creating these masks was crucial. Through signal reformulation and normalization, we achieved 85% sensitivity and 90% specificity in segmenting data into hybrid and pulmonary pixels. This outcome has the potential to enhance the reliability and accuracy of pulmonary perfusion estimation overall.