ABSTRACT
Objective:To investigate the establishment and application value of a radio-mics prediction model for lymph node metastasis of gallbladder carcinoma based on dual-phase enhanced computed tomography (CT).Methods:The retrospective cohort study was conducted. The clinicopathological data of 194 patients with gallbladder carcinoma who were admitted to the First Affiliated Hospital of Xi'an Jiaotong University from January 2012 to December 2020 were collected. There were 70 males and 124 females, aged (64±10)years. All patients underwent curative-intent resection of gallbladder carcinoma. A total of 194 patients were randomly divided into 156 cases in training set and 38 cases in test set according to the ratio of 8:2 based on random number method in R software. The training set was used to establish a diagnostic model, and the test set was used to validate the diagnostic model. After the patients undergoing CT examination, image analysis was performed, radiomics features were extracted, and a radiomics model was established. Based on clinicopathological data, a nomogram prediction model was established. Observation indicators: (1) lymph node dissection and histopathological examination results; (2) establishment and characteristic analysis of a radiomics prediction model; (3) analysis of influencing factors for lymph node metastasis of gallbladder carcinoma; (4) establishment of a nomogram prediction model for lymph node metastasis; (5) comparison of the predictive ability between the radiomics prediction model and nomogram prediction model for lymph node metastasis. Measurement data with normal distribution were represented as Mean± SD, and measurement data with skewed distribution were represented as M(range). Count data were expressed as absolute numbers, and comparison between groups was performed by the chi-square test. Univariate analysis was conducted by the chi-square test, and multivariate analysis was performed by the Logistic regression model forward method. The receiver operating characteristic curve was drawn, and the area under curve, decision curve, confusion matrix were used to evaluate the predictive ability of prediction models. Results:(1) Lymph node dissection and histopathological examination results. Of the 194 patients, 182 cases underwent lymph node dissection, with the number of lymph node dissected as 8(range, 1?34) per person and the number of positive lymph node as 0(range, 0?11) per person. Postoperative histopathological examination results of 194 patients: 122 patients were in stage N0, with the number of lymph node dissected as 7(range, 0?27) per person, 48 patients were in stage N1, with the number of lymph node dissected as 8(range, 2?34) per person and the number of positive lymph node as 1(range, 1?3) per person, 24 patients were in stage N2, with the number of lymph node dissected as 11(range, 2?20) per person and the number of positive lymph node as 5(range, 4?11) per person. (2) Establishment and characteristic analysis of a radiomics prediction model. There were 107 radiomics features extracted from 194 patients, including 18 first-order features, 14 shape features and 75 texture features. According to the intra-group correlation coefficient and absolute median difference of each radiomics feature, mutual information, Select K-Best, least absolute shrinkage and selection operator regression were conducted to further reduce dimensionality. By further combining 5 different machine learning algorithms including random forest, gradient boosting secession tree, support vector machine (SVM), K-Nearest Neighbors and Logistic regression, the result showed that the Select K-Best_SVM model had the best predictive performance after analysis, with the area under receiver operating characteristic curve as 0.76 in the test set. (3) Analysis of influencing factors for lymph node metastasis of gallbladder carcinoma. Results of univariate analysis showed that systemic inflammation response index, carcinoembryonic antigen (CEA), CA19-9, CA125, radiological T staging and radiological lymph node status were related factors for lymph node metastasis of patients with gallbladder cancer ( χ2=4.20, 11.39, 5.68, 11.79, 10.83, 18.58, P<0.05). Results of multivariate analysis showed that carcinoembryonic antigen, CA125, radiological T staging (stage T3 versus stage T1?2, stage T4 versus stage T1?2), radiological lymph node status were independent influencing factors for lymph node metastasis of patients with gallbladder carcinoma [ hazard ratio=2.79, 4.41, 5.62, 5.84, 3.99, 95% confidence interval ( CI) as 1.20?6.47, 1.81?10.74, 1.50?21.01, 1.02?33.31, 1.87?8.55, P<0.05]. (4) Establishment of a nomogram prediction model for lymph node metastasis. A nomogram prediction model was established based on the 4 independent influencing factors for lymph node metastasis of gallbladder carcinoma, including CEA, CA125, radiological T staging and radiological lymph node status. The concordance index of the nomogram model was 0.77 (95% CI as 0.75?0.79) in the training set and 0.73 (95% CI as 0.68?0.72) in the test set, respectively. (5) Comparison of the predictive ability between the radiomics predic-tion model and nomogram prediction model for lymph node metastasis. The receiver operating characteristic curve showed that the areas under the curve of Select K-Best_SVM radiomics model were 0.75 (95% CI as 0.74?0.76) in the training set and 0.76 (95% CI as 0.75?0.78) in the test set, respectively. The areas under the curve of nomogram prediction model were 0.77 (95% CI as 0.76?0.78) in the training set and 0.70 (95% CI as 0.68?0.72) in the test set, respectively. The decision curve analysis showed that Select K-Best_SVM radiomics model and nomogram prediction model had a similar ability to predict lymph node metastasis. The confusion matrix showed that Select K-Best_SVM radiomics model had the sensitivity as 64.29% and 75.00%, the specificity as 73.00% and 59.09% in the training set and test set, respectively. The nomogram had the sensitivity as 51.79% and 50.00%, the specificity as 80.00% and 72.27% in the training set and test set, respectively. Conclusion:A dual-phase enhanced CT imaging radiomics prediction model for lymph node metastasis of gallbladder carcinoma is successfully established, and its predictive ability is good and consistent with that of nomogram.
ABSTRACT
Objective:To investigate the predictive value of clinical radiomics model based on nnU-Net for the prognosis of gallbladder carcinoma (GBC).Methods:The retrospective cohort study was conducted. The clinicopathological data of 168 patients who underwent curative-intent radical resection of GBC in the First Affiliated Hospital of Xi'an Jiaotong University from January 2012 to December 2020 were collected. There were 61 males and 107 females, aged (64±11)years. All the 168 patients who underwent preoperative enhanced computed tomography (CT) examina-tion were randomly divided into 126 cases in training set and 42 cases in test set according to the ratio of 3:1 based on random number table. For the portal venous phase images, 2 radiologists manually delineated the region of interest (ROI), and constructed a nnU-net model to automatically segment the images. The 5-fold cross-validation and Dice similarity coefficient were used to evaluate the generalization ability and predictive performance of the nnU-net model. The Python software (version 3.7.10) and Pyradiomics toolkit (version 3.0.1) were used to extract the radiomics features, the R software (version 4.1.1) was used to screen the radiomics features, and the variance method, Pearson correlation analysis, one-way COX analysis and random survival forest model were used to screen important radiomics features and calculate the Radiomics score (Radscore). X-tile software (version 3.6.1) was used to determine the best cut-off value of Radscore, and COX proportional hazard regression model was used to analyze the independent factors affecting the prognosis of patients. The training set data were imported into R software (version 4.1.1) to construct a clinical radiomics nomogram model of survival prediction for GBC. Based on the Radscore and the independent clinical factors affecting the prognosis of patients, the Radscore risk model and the clinical model for predicting the survival of GBC were constructed respectively. The C-index, calibration plot and decision curve analysis were used to evaluate the predictive ability of different survival prediction models for GBC. Observation indicators: (1) segmentation results of portal venous phase images in CT examination of GBC; (2) radiomic feature screening and Radscore calculation; (3) prognostic factors analysis of patients after curative-intent radical resection of GBC; (4) construction and evaluation of different survival prediction models for GBC. Measurement data with normal distribution were represented by Mean± SD. Count data were expressed as absolute numbers or percentages, and comparison between groups was analyzed using the chi-square test. Univariate and multivariate analyses were performed using the COX proportional hazard regression model. The postoperative overall survival rate was calculated by the life table method. Results:(1) Segmentation results of portal venous phase images in CT examination of GBC: the Dice similarity coefficient of the ROI based on the manual segmentation and nnU-Net segmentation models was 0.92±0.08 in the training set and 0.74±0.15 in the test set, respectively. (2) Radiomic feature screening and Radscore calculation: 1 502 radiomics features were finally extracted from 168 patients. A total of 13 radiomic features (3 shape features and 10 high-order features) were screened by the variance method, Pearson correlation analysis, one-way COX analysis and random survival forest model. Results of random survival forest model analysis and X-tile software analysis showed that the best cut-off values of the Radscore were 6.68 and 25.01. A total of 126 patients in the training set were divided into 41 cases of low-risk (≤6.68), 72 cases of intermediate-risk (>6.68 and <25.01), and 13 cases of high-risk (≥25.01). (3) Prognostic factors analysis of patients after curative-intent radical resection of GBC: the 1-, 2-, and 3-year overall survival rates of 168 patients were 75.8%, 54.9% and 45.7%, respectively. The results of univariate analysis showed that preopera-tive jaundice, serum CA19-9 level, Radscore risk (medium risk and high risk), extent of surgical resection, pathological T staging, pathological N staging, tumor differentiation degree (moderate differentiation and low differentiation) were related factors affecting prognosis of patients in the training set ( hazard ratio=3.28, 3.00, 3.78, 6.34, 4.48, 6.43, 3.35, 7.44, 15.11, 95% confidence interval as 1.91?5.63, 1.76?5.13, 1.76?8.09, 2.49?16.17, 2.30?8.70, 1.57?26.36, 1.96?5.73, 1.02?54.55, 2.04?112.05, P<0.05). Results of multivariate analysis showed that preoperative jaundice, serum CA19-9 level, Radscore risk as high risk and pathological N staging were independent influencing factors for prognosis of patients in the training set ( hazard ratio=2.22, 2.02, 2.89, 2.07, 95% confidence interval as 1.20?4.11, 1.11?3.68, 1.04?8.01, 1.15?3.73, P<0.05). (4) Construction and evaluation of different survival prediction models for GBC. Clinical radiomics model, Radscore risk model and clinical model were established based on the independent influencing factors for prognosis, the C-index of which was 0.775, 0.651 and 0.747 in the training set, and 0.759, 0.633, 0.739 in the test set, respectively. The calibration plots showed that the Radscore risk model, clinical model and clinical radiomics model had good predictive ability for prognosis of patients. The decision curve analysis showed that the prognostic predictive ability of the clinical radiomics model was better than that of the Radscore risk and clinical models. Conclusion:The clinical radiomics model based on the nnU-Net has a good predictive performance for prognosis of GBC.