Effect of Automatic Extraction Accuracy by Different Image Reconstruction Methods Using a Three-dimensional Image Analysis System for Pulmonary Segmentectomy Preoperative CT Angiography

Abstract

This study aimed to determine the optimal image reconstruction method for preoperative computed tomography (CT) angiography for pulmonary segmentectomy. This study enrolled 20 patients who underwent contrast-enhanced CT examination for pulmonary segmentectomy. The optimal image reconstruction algorithm among four different reconstruction algorithms (filtered back projection, hybrid iterative reconstruction, model- based iterative reconstruction, and deep learning reconstruction [DLR]) was investigated by assessing the CT numbers, vessel extraction ratios, and misclassification ratios. The vessel extraction ratios for main and subsegment branches reconstructed using DLR were significantly higher than those using other reconstruction algorithms (96.7% and 90.8% for pulmonary artery and vein, respectively). The misclassification ratios at the right upper lobe pulmonary vessels (V1 and V2) were especially high because they were close to the superior vena cava, and their CT numbers were similar in all four reconstructions. In conclusion, the DLR allows a high extraction rate of pulmonary blood vessels and a low misclassification rate of automatic extraction.