Search

Background: Currently, automatic esophagus segmentation remains a challenging task due to its small size, low contrast, and large shape variation. We aimed to improve the performance of esophagus segmentation in deep learning by applying a strategy that involves locating the object first and then performing the segmentation task. Methods: A total of 100 cases with thoracic computed tomography scans from two publicly available datasets were used in this study. A modified CenterNet, an object location network, was employed to locate the center of the esophagus for each slice. Subsequently, the 3D U-net and 2D U-net_coarse models were trained to segment the esophagus based on the predicted object center. A 2D U-net_fine model was trained based on the updated object center according to the 3D U-net model. The dice similarity coefficient and the 95% Hausdorff distance were used as quantitative evaluation indexes for the delineation performance. The characteristics of the automatically delineated esophageal contours by the 2D U-net and 3D U-net models were summarized. Additionally, the impact of the accuracy of object localization on the delineation performance was analyzed. Finally, the delineation performance in different segments of the esophagus was also summarized. Results: The mean dice coefficient of the 3D U-net, 2D U-net_coarse, and 2D U-net_fine models were 0.77, 0.81, and 0.82, respectively. The 95% Hausdorff distance for the above models was 6.55, 3.57, and 3.76, respectively. Compared with the 2D U-net, the 3D U-net has a lower incidence of delineating wrong objects and a higher incidence of missing objects. After using the fine object center, the average dice coefficient was improved by 5.5% in the cases with a dice coefficient less than 0.75, while that value was only 0.3% in the cases with a dice coefficient greater than 0.75. The dice coefficients were lower for the esophagus between the orifice of the inferior and the pulmonary bifurcation compared with the other regions. Conclusion: The 3D U-net model tended to delineate fewer incorrect objects but also miss more objects. Two-stage strategy with accurate object location could enhance the robustness of the segmentation model and significantly improve the esophageal delineation performance, especially for cases with poor delineation results. [ABSTRACT FROM AUTHOR]

Objective: We conducted a retrospective statistical analysis of clinical and physical dosimetric factors of lung cancer patients who had previously undergone lung and/or mediastinal radiotherapy and died of or survived severe acute radiation pneumonitis (SARP). Our study was the first to reveal the heterogeneity in clinical factors, physical dosimetric factors, and SARP onset time that determined the clinical outcomes of lung cancer patients who developed SARP. Materials and methods: The clinical characteristics, physical dosimetry factors, and SARP onset time of deceased and surviving patients were retrospectively analyzed. SPSS 20.0 was used for data analysis. Student's t-test was used for intergroup comparison, and a Mann–Whitney U test was used for data with skewed distribution. Qualitative data were represented using frequencies (%), and Fisher's exact test or χ2 test was used for intergroup comparison of nonparametric data. Binary logistic analysis was used for univariate and multivariate analyses. Differences with a P < 0.05 were considered statistically significant. Results: Univariate analysis revealed that the potential predictors of SARP death were as follows: ipsilateral lung V5 and V30, contralateral lung V5, V10, and V30, total lung V5, V10, and V30, mean lung dose, mean heart dose, and maximum spinal cord dose. Multivariate analysis showed that ipsilateral lung V5 and total lung V5 were predictors that determined the final outcome of SARP patients. In addition, we analyzed the time from the completion of radiotherapy to SARP onset, and found significant difference between the two groups. Conclusions: There was no decisive correlation between clinical characteristics and SARP outcome (i.e., death or survival) in lung radiotherapy patients. Ipsilateral lung V5 and total lung V5 were independent predictors of death in SARP patients. [ABSTRACT FROM AUTHOR]