Publications by authors named "Eduardo Horneaux de Moura"
Article Synopsis
- Endoscopic ultrasound (EUS) is an important tool for evaluating pancreatic lesions, but its accuracy is not very high, prompting the development of a convolutional neural network (CNN) to improve diagnosis of pancreatic cystic neoplasms (PCN) and pancreatic solid lesions (PSL).
- The CNN was trained on a large dataset from four international centers, comprising over 126,000 images, and successfully distinguished between different types of pancreatic lesions with high accuracy.
- The results showed the CNN achieved over 99% accuracy for normal pancreatic tissue and PCNs, while distinguishing between pancreatic ductal adenocarcinoma (P-DAC) and neuroendocrine tumors (P-NET) at 94% accuracy, pointing to the