Manuscript received June 4, 2025; revised July 8, 2025; accepted August 20, 2025; published December 19, 2025.

Abstract—Pancreatic Cancer (PC) is a very typical to treat and cure among different type of cancers, which is often found too late and spreads quickly, making survival rates very low. Imaging modalities has mostly been used to diagnose pancreatic cancer, but the most up-to-date imaging gives a bad outlook, which limits the treatments doctors can use. By combining deep learning and machine learning, doctors can make better decisions and find cancer earlier. In this paper, DesnseNet 201 a deep learning model and Support Vector Machine (SVM) classifier is utilized for identification of PC. Three stage processing of input data is performed to achieve the results. First stage is data preprocessing; second stage is extraction of features using DenseNet 201 model and reduced using Principal Component Analysis (PCA), final stage is classification using SVM model. The most significant features from the given input image are been extracted using the process of DenseNet 201 model and finalized using PCA before fetching to classifier. The parameters like accuracy, precision, recall, Specificity and F-measure are evaluated. The accuracy obtained using the proposed three stage detection model is 95.52% and is better compared to traditional model. The F1-Score achieved is 94.14 and Area Under the Curve (AUC) is 96.5%.

Keywords—pancreatic cancer, Computed Tomography (CT) images, principal component analysis, DenseNet 201, Support Vector Machine (SVM)

Cite: Anagani V. Shanthi, Alli D. Rani, Panuganti Ravi, and Muppalla Tharangini, "DenseNet-SVM: An Intelligent Model for Pancreatic Cancer Detection," Journal of Image and Graphics, Vol. 13, No. 6, pp. 673-685, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.