The aim of this study was to assess the diagnostic ability of radiomics combined with multiple machine learning algorithms to differentiate pancreatic ductal adenocarcinoma (PDAC) from pancreatic neuroendocrine tumor (pNET). This retrospective study included a total of 238 patients diagnosed with PDAC or pNET. Using specialized software, radiologists manually mapped regions of interest (ROIs) from computed tomography images and automatically extracted radiomics features. A total of 45 discriminative models were built by five selection algorithms and nine classification algorithms. The performances of the discriminative models were assessed by sensitivity, specificity and the area under receiver operating characteristic curve (AUC) in the training and validation datasets. Using the combination of Gradient Boosting Decision Tree (GBDT) as the selection algorithm and Random Forest (RF) as the classification algorithm, the optimal diagnostic ability with the highest AUC was presented in the training and validation datasets. The sensitivity, specificity and AUC of the model were 0.804, 0.973 and 0.971 in the training dataset and 0.742, 0.934 and 0.930 in the validation dataset, respectively. The combination of radiomics and multiple machine learning algorithms showed the potential ability to discriminate PDAC from pNET. We suggest that multi-algorithm modeling should be considered for similar studies in the future rather than using a single algorithm empirically.
Purpose: The purpose of our study was to investigate the clinical significance and prognostic role of the systemic immune‐inflammation index (SII) in patients who underwent surgical resection for nonfunctioning pancreatic neuroendocrine tumors (pNETs).Methods: We conducted a retrospective analysis of 364 patients with nonfunctioning pNETs. The association between the SII level and clinical parameters was investigated. The receiver operating characteristic (ROC) curve was used to calculate the optimal SII value. Cox proportional hazard analysis was performed to evaluate the prognostic factors.Results: Our study included 364 patients with nonfunctioning pNETs who underwent surgery. The median age was 51.0 (43.0, 59.3), and 164 (45.1%) were male. The optimal threshold of SII determined by ROC analysis was 523.95. Higher SII levels were significantly associated with older age (p = 0.001), sex (p = 0.011), tumor size (p = 0.032), and tumor grade (p = 0.002). Recurrence was observed in 70 (19.2%) patients following a median follow‐up of 98 months. Univariate analysis showed that higher SII (p < 0.0001), tumor size >4 cm (p = 0.015), and G2/G3 grade (p = 0.002) were significantly associated with disease‐free survival (DFS). Multivariate analysis revealed that higher SII (HR: 7.35; 95% CI: 3.44, 15.70; p < 0.0001) and G2/G3 grade (HR: 3.11; 95% CI: 1.42, 6.82; p = 0.005) remained significantly associated with tumor recurrence. Furthermore, 46 (12.6%) patients died during the follow‐up. Higher SII (HR: 8.43; 95% CI: 3.19, 22.72; p < 0.0001) and G2/G3 grade (HR: 3.16; 95% CI: 1.01, 9.86; p = 0.048) were independent predictors of overall survival (OS) by multivariate analysis.Conclusion: In conclusion, our study revealed that a higher SII level was associated with tumor‐related features (larger tumor size and advanced grade) and subsequent shorter DFS and OS in patients with nonfunctioning pNETs. These results indicated that the SII could serve as an efficient prognostic biomarker for nonfunctioning pNETs.