Pancreatic cancer, one of the worst cancers in the world, It has a poor five-year survival rate. Although early diagnosis with medical imaging scans has improved outcomes, accurately identifying the types of tumors remains challenging, especially when it comes to pancreatic neuroendocrine tumors (PNETs). Even with its widespread use, magnetic resonance imaging (MRI) is not always accurate, and the process of manually identifying tumor anomalies is laborious. Advances in deep learning and optimization techniques yield more efficient solutions that outperform traditional methods in terms of diagnostic accuracy. It has been demonstrated that the accuracy of pancreatic tumor identification with Hierarchical Neural Networks (HCNN) employing convolutional filters for pattern recognition with Internet of Things sensors is 97.5%. The existing models perform well, with a 96% accuracy rate; however, future research will focus on reducing errors and increasing precision.