Machine learning via DARTS-Optimized MobileViT models for pancreatic Cancer diagnosis with graph-based deep learning - Research Open Access

Sep 23, 2026

Uploaded by Aria

Pancreatic cancer diagnosis is challenging because the disease is often asymptomatic and detected at an advanced stage. This study proposes a graph-based learning pipeline that transforms pancreatic CT images into graph structures using the Harris Corner Detection algorithm to model complex spatial relationships. Graph representations are then processed with MobileViT models optimized via Differentiable Architecture Search (DARTS) for dynamic architecture adaptation. Classical machine-learning classifiers (KNN, SVM, RF, XGBoost) are applied to improve classification, with MobileViTv2_150 and MobileViTv2_200 reaching 97.33% accuracy and 96.25% F1, outperforming CNN and ViT baselines, and supporting scalable clinical imaging applications.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages21

Reading Time53 min

AccessFree to download

Abstract

Highlights

Keywords

Graphical Abstract

Introduction

Alaca BMC Medical Informatics and Decision Making [https://doi.org/10.1186/s12911-025-02923-x](https://doi.org/10.1186/s12911-025-02923-x) (2025) 25:81 BMC Medical Informatics and Decision Making RESEARCH Open Access Machine learning via DARTS-Optimized MobileViT models for pancreatic Cancer diagnosis with graph-based deep learning Yusuf Alaca1* Abstract The diagnosis of pancreatic cancer presents a significant challenge due to the asymptomatic nature of the disease and the fact that it is frequently detected at an advanced stage. This study presents a novel approach combining graph-based data representation with DARTS-optimised MobileViT models, with the objective of enhancing diagnostic accuracy and reliability. The images of the pancreatic CT were transformed into graph structures using the Harris Corner Detection algorithm, which enables the capture of complex spatial relationships. Subsequently, the graph representations were processed using MobileViT models that had been optimised with Differentiable Architecture Search (DARTS), thereby enabling dynamic architectural adaptation. To further enhance classification accuracy, advanced machine learning algorithms, including K-Nearest Neighbours (KNN), Support Vector Machines (SVM), Random Forest (RF), and XGBoost, were applied. The MobileViTv2_ 150 and MobileViTv2_200 models demonstrated remarkable performance, with an accuracy of 97. 33% and an F1 score of 96 . 25%, surpassing the capabilities of traditional CNN and Vision Transformer models. This innovative integration of graph-based deep learning and machine learning techniques demonstrates the potential of the proposed method to establish a new standard for early pancreatic cancer diagnosis. Furthermore, the study highlights the scalability of this approach for broader applications in medical imaging, which could lead to improved patient outcomes. Highlights • The performance of MobileViT models for pancreatic cancer diagnosis was enhanced using DARTS for dynamic optimization. • By converting pancreatic CT images into graph representations, complex structural features can be captured, thereby enhancing diagnostic accuracy. • MobileViT models have been integrated by combining the strengths of CNN architectures and visual transformer (ViT) models in medical imaging. • Classification algorithms have been utilized to improve the classification performance of optimized MobileViT models, achieving superior accuracy and reliability. • Comparative analyses with traditional CNNs and Vision Transformer models highlight the advantages ofthe proposed approach. • This approach has the potential to improve patient outcomes and survival rates by enhancing early pancreatic cancer diagnosis. *Correspondence: Yusuf Alaca [yusufalaca@hitit.edu.tr](yusufalaca@hitit.edu.tr) © The Author(s) 2025. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/.)[.](http://creativecommons.org/licenses/by-nc-nd/4.0/.)

FAQs

Why is pancreatic cancer diagnosis difficult in early stages?

Early diagnosis is difficult due to nonspecific or absent symptoms and frequent detection at a late stage, which leads to poor prognoses.

How are CT images converted for the proposed model?

Pancreatic CT images are transformed into graph structures using the Harris Corner Detection algorithm to capture complex spatial relationships.

What performance did the optimized MobileViT models achieve?

MobileViTv2_150 and MobileViTv2_200 achieved 97.33% accuracy and 96.25% F1 score, exceeding traditional CNN and Vision Transformer approaches.

Sep 23, 2026

Uploaded by Aria