Detection and Classification of Pancreatic Cancer in CT Scan Images Using a Neural Network Approach Based on Inception V۳

The lack of specific symptoms in the early stages of pancreatic cancer has made this cancer one of the most dangerous cancers. Effective and timely screening is not possible due to the lack of early symptoms. CT scan images are widely used for clinical examinations. The diagnosis and classification of pancreatic cancer using medical image analysis is a new and vital field. This cancer is examined with CT scan images, which are in most cases without contrast. In this study, while overcoming this challenge, a method for the diagnosis and classification of pancreatic cancer is introduced. In this study, our goal is to diagnose and classify pancreatic cancer using a deep learning method, which is not possible to detect this hidden cancer with the naked eye. This network provides information with classification capabilities by extracting features, and the information obtained can help determine the phase of the disease and timely treatment of the patient. The main basis of our classification and diagnosis model is optimized on the Inception v۳ network. Extensive experiments have been conducted to evaluate the proposed technique on a database of pancreatic images. The simulation results show a maximum sensitivity of ۹۹.۲۵%, specificity of ۹۹.۲۸%, and accuracy of ۹۹.۳۱%

کلیدواژه ها:

neural network ، Pancreatic cancer ، early detection ، deep learning ، inception v۳