Evaluation of Artificial Intelligence in Defining Lung Ultrasound Findings in Anesthesia Management of Preeclampsia

A total of eight quadrants of standard lung US examination was performed to all pregnant women with preeclampsia within the scope of the study by the same anesthesiologist, dividing each hemithorax into four regions through the parasternal, anterior axillary, posterior axillary vertical lines, and the horizontal line assumed to pass under the nipple. The resulting images were stored in digital media. A Lung US examination was performed once for an average of 5-10 minutes. The presence of B lines was investigated in the examination.

The case was defined as interstitial edema when the B lines, which are defined as vertical linear hyperechoic reverberation artifacts representing the edematous interlobular septa/alveoli, extend posteriorly from below the pleural line and moving in sync with lung movements, are found in two or more lung areas. B-lines were determined for each case and reported in standard form in terms of number and morphology. The diagnostic accuracy of B lines was determined with the artificial intelligence supported SmartAlpha Rievi 1300 software program. B-lines validated by artificial intelligence assisted algorithm in all stored digital images were reported blindly by another anesthesiologist experienced in lung US. The clinical features, laboratory parameters, and intraoperative hemodynamic data of the cases were recorded to be evaluated in terms of relationship with lung US data.

We predict that the application of lung US with artificial intelligence software will provide an opportunity to quickly evaluate the clinic of preeclamptic pregnant women who are frequently operated on in emergency conditions.