Multi-Objective Genetic Algorithm Optimised Convolutional Neural Networks for Improved Pancreatic Cancer Detection
S. Kamatchi, S. Preethi, K. Suresh Kumar, N Nandini, Desidi Narsimha Reddy, S. Karthick · 2025
Pancreatic tumour is the deadliest disease and require early detection to reduce mortality. Advanced imaging techniques have improved diagnostic accuracy, however identifying and classifying pancreatic tumors requires robust methods to assist radiologists and clinicians in the diagnostic process. In light of this, the research uses Multi-Objective Genetic Algorithm (MOGA) optimized Convolutional Neural Networks (CNN) for the diagnosis of pancreatic tumours. The process begins with image preprocessing utilizing Wiener filter to reduce noise, followed by segmentation through thresholding to separate region of interest. After, feature extraction in then performed using Gray level co-occurrence matrix (GLCM), a texture analysis technique that captures spatial relationship within image. Finally, classification is accomplished using CNN optimized by a MOGA. The objective of the proposed optimized approach supports in improving the accuracy providing reliable model for early detection and classification of pancreatic tumours. The examination of the proposed work is done using Python with predicting pancreatic cancer dataset and the validation outcomes reveals the effectiveness of proposed model with classification of accuracy of 93.9% and the model loss is 0.35.