Deep Learning Algorithm for Early Pancreatic Cancer Diagnosis

Abstract

Pancreatic cancer is usually detected late, making treatment difficult and survival rates low. Early detection using biomarkers is important, but no reliable biomarker has yet reached clinical use due to challenges in sample collection and the highly variable nature of pancreatic tumors. To improve diagnosis, new machine learning and deep learning methods are proposed. The research focuses on segmenting and classifying MRI/CT images and improving classifier performance using medical data. The proposed deep-learning model (HdiGTF-SIRNN) significantly improves accuracy and precision compared to existing methods, and additional techniques (RIDT-GDLBC and DHEGQDRLCS) also show better diagnostic performance.