Retrospective study of BRAFV600E mutation and CT features of papillary thyroid carcinoma.

Objective: This study aimed to examine the correlation between BRAFV600E status and computed tomography (CT) imaging characteristics in papillary thyroid carcinoma (PTC) and determine if suspicious CT imaging features could predict BRAFV600E status. Methods: This retrospective study included patients with pathologically confirmed PTC at the Department of Thyroid Surgery of Zhongshan Hospital, Xiamen University, between July 2020 and June 2022. We compared the clinicopathologic factors and CT findings of nodules with and without the mutation, and the multiple logistical regression test was used to determine independent parameters of the BRAFV600E mutation. Results: This study included 381 patients with PTC, among them, BRAFV600E mutation was detected in 314 patients (82.4%). Multivariate logistic regression analysis showed that gender (OR = 0.542, 95% CI [0.296-0.993], P = 0.047) and shape (OR = 0.510, 95% CI [0.275-0.944], P = 0.032) were associated with BRAFV600E mutation. Conclusions: Compared to BRAFV600E mutation-negative, BRAFV600E-positive PTC lesions were more likely to be found in female patients and were characterized by irregular shape. However, the CT imaging finding is not enough to predict BRAFV600E status, but an indication.

Comparative analysis of automatic segmentation of esophageal cancer using 3D Res-UNet on conventional and 40-keV virtual mono-energetic CT Images: a retrospective study.

Objectives: To assess the performance of 3D Res-UNet for fully automated segmentation of esophageal cancer (EC) and compare the segmentation accuracy between conventional images (CI) and 40-keV virtual mono-energetic images (VMI40 kev). Methods: Patients underwent spectral CT scanning and diagnosed of EC by operation or gastroscope biopsy in our hospital from 2019 to 2020 were analyzed retrospectively. All artery spectral base images were transferred to the dedicated workstation to generate VMI40 kev and CI. The segmentation model of EC was constructed by 3D Res-UNet neural network in VMI40 kev and CI, respectively. After optimization training, the Dice similarity coefficient (DSC), overlap (IOU), average symmetrical surface distance (ASSD) and 95% Hausdorff distance (HD_95) of EC at pixel level were tested and calculated in the test set. The paired rank sum test was used to compare the results of VMI40 kev and CI. Results: A total of 160 patients were included in the analysis and randomly divided into the training dataset (104 patients), validation dataset (26 patients) and test dataset (30 patients). VMI40 kevas input data in the training dataset resulted in higher model performance in the test dataset in comparison with using CI as input data (DSC:0.875 vs 0.859, IOU: 0.777 vs 0.755, ASSD:0.911 vs 0.981, HD_95: 4.41 vs 6.23, all p-value <0.05). Conclusion: Fully automated segmentation of EC with 3D Res-UNet has high accuracy and clinically feasibility for both CI and VMI40 kev. Compared with CI, VMI40 kev indicated slightly higher accuracy in this test dataset.