Nomogram based on prognostic nutrition index and Chest CT imaging signs predicts lymph node metastasis in NSCLC patients

OBJECTIVE

To establish and validate a model capable of predicting lymph node metastasis (LNM) of non-small cell lung cancer (NSCLC) patients.

METHODS

Preoperative clinical and CT imaging data on patients with NSCLC undergoing surgery were retrospectively analyzed. A model was developed using a training cohort of 290 patients. The univariate analysis followed by dichotomous logistic regression was performed to estimate different risk factors of lymph node metastasis, and a nomogram was constructed. Using another testing cohort of 120 patients, the performance of the nomogram was validated using several evaluation methods and indices and evaluated including via the area under the curve (AUC), calibration curve, Hosmer-Lemeshow test and decision curve analysis (DCA).

RESULTS

CT-based imaging signs were important independent risk factors for lymph node metastasis in NSCLC patients. The possible risk factors also included four other independent risk factors through dichotomous logistic regression, i.e., age, SIRI, PNI and CEA, which were filtered and included in the nomogram. Nomogram yields AUC values of 0.828 [95% confidence interval (CI): 0.778-0.877] in the training cohort and 0.816 (95% CI: 0.737-0.895) in the validation cohort, respectively. The calibration curves showed high agreement in both the training and validation cohorts. At the threshold probability of 0-0.8, the nomogram increases the net outcomes compared to the treat-none and treat-all lines in the decision curve.

CONCLUSIONS

The nomogram based on the PNI and CT images signs holds promise as a novel and accurate tool for predicting the LNM in NSCLC patients and guiding intraoperative lymph node dissection.

A radiomics model of liver CT to predict risk of hepatic encephalopathy secondary to hepatitis B related cirrhosis

PURPOSE

To build a radiomics model of liver contrast-enhanced computed tomography (CT) to predict hepatic encephalopathy secondary to Hepatitis B related cirrhosis.

MATERIALS AND METHODS

This study consisted of 304 consecutive patients with first-diagnosed hepatitis B related cirrhosis. 212 and 92 patients were randomly computer-generated into training and testing cohorts, among which 38 and 21 patients endured HE, respectively. 356 radiomics features of liver were extracted from portal venous-phase CT data, and 3 clinical features were collected from medical record. After data were standardized by Z-score, we used least absolute shrinkage and selection operator to choose useful radiomics features. Ultimately, three predictive models including a radiomics model, a clinical model and an integrated model of radiomics and clinical features were built by analysis of R-software. Predictive performance was tested by multivariable logistic regression, and evaluated by area under receiver-operating characteristic curve (AUC), and accuracy.

RESULTS

19 radiomics features of liver CT were selected. The selected radiomics features and 3 relevant clinical features were applied to develop a radiomics model, a clinical model, and an integrated model of both radiomics and clinical features. The integrated model showed better performance than the radiomics model or clinical model to predict HE (AUC = 0.94 vs. 0.91 or 0.76, and 0.87 vs. 0.86 or 0.73; accuracy = 0.93 vs. 0.89 or 0.83, and 0.83 vs. 0.84 or 0.77) in the training and testing cohorts, respectively.

CONCLUSION

The integrated model of radiomics and clinical features could well predict HE secondary to hepatitis B related cirrhosis.

Predicting gastro-oesophageal variceal bleeding in hepatitis B-related cirrhosis by CT radiomics signature

AIM

To develop liver a computed tomography (CT) radiomics model to predict gastro-oesophageal variceal bleeding (GVB) secondary to hepatitis B-related cirrhosis.

MATERIALS AND METHODS

Electronic medical records and image data of liver triple-phase contrast-enhanced CT examinations of 295 patients with hepatitis B-related cirrhosis were collected retrospectively from two hospitals. Two hundred and thirty-six and 59 patients were enrolled randomly into the training and validation cohorts, respectively; and 75 in the training cohort and 16 in the validation cohort endured GVB while the others did not during follow-up period. Radiomics features of the liver were extracted from the portal venous phase images, and clinical features came from medical records. The tree-based method and univariate feature selection were used to select useful features. The radiomics model, clinical model, and integration of radiomics and clinical models were built using the useful image features and/or clinical features. Predicting performance of three models was evaluated with the area under receiver-operating characteristic curve (AUC), accuracy, and F-1 score.

RESULTS

Twenty-one useful radiomics features and/or three clinical features were selected to build prediction models that correlated with GVB. AUC of integration of radiomics and clinical models was larger than of clinical or radiomics models for the training cohort (0.83±0.09 versus 0.64±0.08 or 0.82±0.10) and the validation cohort (0.64 versus 0.61 or 0.61). Integration of radiomics and clinical models obtained good performance in predicting GVB for both the training and validation cohorts (accuracy: 0.76±0.07 and 0.73, and F-1 score: 0.77±0.09 and 0.72, respectively).

CONCLUSION

Integration of the radiomics and clinical models may be a non-invasive method to predict GVB.