Integrative Deep Diagnostic Architecture: A Convolutional Variational Attention Transformer for Personalized and Multi-Strategy Lung Nodule Detection

Read full text Summarise with VIRAT
Authors: Zia ur Rehman, Juanjuan Zhao, Muhammad Zaman, Qiang Yan, Aoyu Li, Adnan Akhunzada
Cited by: 0
References: 50

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Lung cancer persists as a primary cause of cancer mortality worldwide, largely due to delayed diagnosis and the morphological similarity between benign and malignant pulmonary nodules in early imaging. To enhance early diagnostic precision, this study introduces a Three-Dimensional Convolutional Variational Attention Network (3D-ConVAT), a generative deep learning framework that unifies learning of convolutional representation, variational encoding, and contextual reasoning driven by attention. The proposed architecture integrates a Faster VGG16 detection head with a SegNet-inspired encoder-decoder backbone, enabling precise volumetric nodule localisation and robust classification through latent-space feature embedding. A Support Vector Machine (SVM) classifier refines decision boundaries, distinguishing malignancy with increased reliability. To promote personalized clinical interpretation, physiological biomarkers and patient-specific indicators are fused within the decision process, bridging computational inference with the biomedical context. Moreover, integration with IoT-enabled Wireless Body Area Networks (WBANs) supports continuous physiological monitoring, positioning the framework within the broader domain of AI-driven personalized healthcare electronics. Experimental validation in the LIDC-IDRI dataset demonstrates a sensitivity of 97.2%, a specificity of 95.4% and superior MCC and AUC scores compared to contemporary methods. The results highlight the diagnostic precision, interpretability, and computational efficiency of the system, underscoring its potential for next-generation intelligent, personalised, and wearable-assisted lung cancer detection and monitoring.

Fields of Study & Subjects

AI in cancer detection Lung Cancer Diagnosis and Treatment Radiomics and Machine Learning in Medical Imaging

Publication details

Journal / source
IEEE Access
Volume
14
Pages
52054–52066
Publication date
1 January 2026
Language
EN