Predicting Cancer in Pancreatic Cystic Lesions Through Artificial Intelligence

This international, multicenter retrospective study aims to develop a deep learning (DL)-based predictive model to identify malignant transformation in pancreatic cystic lesions, improving upon current clinical guidelines. The model will integrate clinical, biochemical, and multimodal imaging data. Several 3D convolutional neural networks will be trained using advanced preprocessing, data augmentation, and hybrid fusion techniques. Model performance will be compared to that of existing international guidelines. The study involves no additional procedures for patients and adheres to strict data anonymization and privacy regulations.