RESUMEN
Objective:To investigate the influencing factors for celiac lymph node metastasis in thoracic esophageal squamous cell carcinoma (TE-SCC), construct a prediction model of celiac lymph node metastasis in TE-SCC, and stratify the probability of celiac lymph node metastasis.Methods:The retrospective case-control study was conducted. The clinicopathological data of 443 patients with TE-SCC who underwent thoracoscopic and laparoscopic esophagectomy with systematic lymph node dissection in the First Affiliated Hospital of Zhengzhou University between March 2015 and April 2019 were collected. There were 259 males and 184 females, aged from 41 to 81 years, with a median age of 64 years. The nomogram prediction model was constructed based on the results of multivariate analysis of influencing factors for celiac lymph node metastasis in TE-SCC, of which calibration curve and decision curve were drawed. The predictive performance was evaluated using the concordance index. The score for celiac lymph node metastasis in TE-SCC predicted by nomogram model was used for further recursive partitioning analysis, and patients were stratified into risk subgroups using the decision-making tree model. Observation indicators: (1) celiac lymph node metastasis in TE-SCC; (2) analysis of influencing factors for celiac lymph node metastasis in TE-SCC; (3) construction of nomogram prediction model of celiac lymph node metastasis in TE-SCC; (4) construction of decision-making tree model of celiac lymph node metastasis in TE-SCC and risk subgroup analysis of celiac lymph node metastasis probability. Measurement data with skewed distribution were represented as M (range). Count data were represented as absolute numbers and percentages, and comparison between groups was analyzed using the chi-square test. Comparison of ordinal data between groups was analyzed using the nonparametric rank sum test. Multivariate analysis was performed using the Logistic regression model. Based on Logistic regression model multivariate analysis, a new nomogram model was constructed using the RStudio 3.4 software. Results:(1) Celiac lymph node metastasis in TE-SCC: celiac lymph node metastasis was found in 89 of the 443 patients, with a celiac lymph node metastasis rate of 20.09%(89/443). (2) Analysis of influencing factors for celiac lymph node metastasis in TE-SCC. Results of univariate analysis showed that tumor location, tumor length, tumor differentiation degree, pathological T staging, nerve invasion, vessel invasion, and thoracic lymph node metastasis were related factors for celiac lymph node metastasis in TE-SCC ( χ2=12.177, Z=-2.754, -4.218, -4.254, χ2=3.908, 33.025, 30.387, P<0.05). Results of multivariate analysis showed that tumor location, vessel invasion, and thoracic lymph node metastasis were independent influencing factors for celiac lymph node metastasis in TE-SCC ( odds ratio=2.165, 3.442, 2.876, 95% confidence interval: 1.380-3.396, 1.787-6.633, 1.631-5.071, P<0.05). (3) Construction of nomogram prediction model of celiac lymph node metastasis in TE-SCC: based on the factors screened by multivariate analysis, including tumor location, vessel invasion, and thoracic lymph node metastasis, the nomogram prediction model of celiac lymph node metastasis in TE-SCC was established, with the concordance index of 0.846. The calibration curve showed a high consistency between the celiac lymph node metastasis probability estimated by the prediction model and the actual rate of celiac lymph node metastasis. The decision curve showed that the nomogram prediction model of celiac lymph node metastasis in TE-SCC had a good prediction value when the probability threshold was 0.001-0.819.(4) Construction of decision-making tree model of celiac lymph node metastasis in TE-SCC and risk subgroup analysis of celiac lymph node metastasis probability: patients were stratified into six risk subgroups using the decision-making tree model based on the celiac lymph node metastasis probability. The group A included patients with no vessel invasion+negative thoracic lymph node, group B included patients with no vessel invasion+the number of positive thoracic lymph nodes of 1-3, group C included patients with no vessel invasion+the number of positive thoracic lymph nodes of ≥4, group D included patients with vessel invasion+the number of positive thoracic lymph nodes of 0-2+upper or middle thoracic esophageal carcinoma, group E included patients with vessel invasion+the number of positive thoracic lymph nodes of 0-2+lower thoracic esophageal carcinoma, group F included patients with vessel invasion+the number of positive thoracic lymph nodes of ≥3. The group A was low-risk group with the celiac lymph node metastasis probability of 11%, group B and D were intermediate low-risk groups with the celiac lymph node metastasis probability of 27% and 21%, group C and E were the intermediate high-risk groups with the celiac lymph node metastasis probability of 56% and 55%, and group F was high-risk group with the celiac lymph node metastasis probability of 80%. Conclusions:The tumor location, vessel invasion, and thoracic lymph node metastasis are independent influencing factors for celiac lymph node metastasis in TE-SCC. Vessel invasion has the dominant influence on celiac lymph node metastasis, followed by the number of positive thoracic lymph nodes, and then the tumor location. Patients can be stratified into six risk subgroups based on the nomogram prediction model and decision-making tree model of celiac lymph node metastasis in TE-SCC.
RESUMEN
A method was proposed to detect pulmonary nodules in low-dose computed tomography (CT) images by two-dimensional convolutional neural network under the condition of fine image preprocessing. Firstly, CT image preprocessing was carried out by image clipping, normalization and other algorithms. Then the positive samples were expanded to balance the number of positive and negative samples in convolutional neural network. Finally, the model with the best performance was obtained by training two-dimensional convolutional neural network and constantly optimizing network parameters. The model was evaluated in Lung Nodule Analysis 2016(LUNA16) dataset by means of five-fold cross validation, and each group's average model experiment results were obtained with the final accuracy of 92.3%, sensitivity of 92.1% and specificity of 92.6%.Compared with other existing automatic detection and classification methods for pulmonary nodules, all indexes were improved. Subsequently, the model perturbation experiment was carried out on this basis. The experimental results showed that the model is stable and has certain anti-interference ability, which could effectively identify pulmonary nodules and provide auxiliary diagnostic advice for early screening of lung cancer.