ABSTRACT Pancreatic cancer is a type of cancer that is difficult to diagnose and is often detected at later stages. The lack of symptoms in the early stages and the limitations of current diagnostic methods reduce the treatability rate of the disease. Therefore, early detection of pancreatic cancer is vital to increasing patients' lifespan and quality of life. In this study, a comprehensive model combining deep learning and optimization techniques using high‐resolution CT images for the diagnosis of pancreatic cancer is proposed. The proposed model aims to significantly enhance diagnostic accuracy and efficiency by integrating the MobileNetV2 Convolutional Neural Network (CNN) and the Ant Lion Optimization (ALO) algorithm. Initially, CT images of normal and pancreatic tumor cases were processed using the Harris edge detection algorithm to highlight important structural features. These images were then converted into a graph structure where each pixel is represented as a node and neighboring pixels as edges. The graph images were classified using the MobileNetV2 CNN model, known for its low computational cost and high performance. Following classification, deep features were extracted and optimized using the ALO algorithm. ALO is an optimization algorithm based on the hunting strategies of ant lions. The optimized features were classified using various machine learning algorithms such as Decision Trees (DT), k‐Nearest Neighbors (kNN), Naive Bayes (NB), and Support Vector Machines (SVM). Among these algorithms, SVM achieved the highest accuracy, making it the most effective algorithm for pancreatic cancer diagnosis. To enhance the model's robustness and generalizability, a 5‐fold cross‐validation technique was used, proving the model's capacity to provide reliable and accurate diagnostics. This model offers a promising method that supports the early detection of pancreatic cancer by improving diagnostic accuracy in clinical applications.
Alaca et al. (2026) studied this question.