Preoperative Prediction of Pancreatic Neuroendocrine Neoplasms Grading Based on Enhanced CT Imaging: Validation of Deep Learning with a Convolutional Neural Network

Abstract

Background: The pathological grading of pancreatic neuroendocrine neoplasms (pNENs) is an independent predictor of survival impact and treatment protocol. Deep learning (DL) may improve the preoperative prediction of pNEN grading. Methods: Ninety-three pNEN patients with preoperative CECT scanning from our institution were retrospectively enrolled. A CNN-based DL algorithm was applied to the CECT images to obtain three models (arterial, venous, and arterial/venous models), the performances of which were evaluated via an eight-fold cross-validation technique. The CECT images of the optimal phase were used for comparing the DL and traditional machine learning (TML) models in predicting the pathological grading of pNENs. The performance of two radiologists was also evaluated. The best DL model from the eight-fold cross-validation was evaluated on an independent testing set of 19 patients from another institution who were scanned on a different scanner. Findings: The AUC (0.81) of arterial phase was significantly higher than those of venous (P = 0.03) and arterial/venous phase (P = 0.03) in predicting the pathological grading of pNENs. Compared with the TML models, the DL model gave a higher (although insignificantly) AUC. The pathological grading in 54 of the 93 patients (58.1%) was correctly assessed by the radiologists. The DL algorithm achieved an AUC of 0.82 and an accuracy of 88.1% for the independent testing set. Interpretation: The CNN-based DL showed a relatively robust performance in predicting pathological grading of pNENs from CECT images. Funding Statement:This work was funded by National Natural Science Foundation of China (81771908, 81571750, 81770654, 81801761, 81471735), National Key Research and Development Program of China (2017YFC0113402), Guangzhou Science and Technology Foundation (201804010078). Declaration of Interests: The authors declare no conflicts of interest. Ethics Approval Statement: The protocol for this retrospective study was approved by our institutional review board. The procedures carried out in the study were conducted by adhering to the principles of the Declaration of Helsinki. A written informed consent was obtained from each patient.