Scholarlink

<p>Breast cancer remains a leading cause of mortality among women globally, and early, accurate diagnosis is essential for effective treatment. Integrating IoT with intelligent diagnostic systems enables efficient and remote healthcare delivery. Traditional convolution-based diagnostic models often fail to capture the spatial and relational intricacies in mammographic images and lack scalability for real-time use. To address this, we propose a novel IoT-enabled framework leveraging the visual geometry-grounded non-convolutional graph neural network (VG2NCGNN), which incorporates observability-constrained resampling-free cubature Kalman filter for noise-resistant preprocessing, inverse Z-transform, and Wiener–Hopf factorization for advanced feature extraction, and Visual Geometry-Grounded Transformers for spatial reasoning. The model is optimized using the Addax Optimization Algorithm (AOA) to enhance accuracy and convergence. Designed for real-time deployment with IoT-connected mammography scanners, the VG2NCGNN-AOA framework demonstrated superior performance on the BreakHis and MIAS datasets, achieving 99.2% accuracy, 98.6% precision, 98.1% sensitivity, 98.3% F1-score, 98.9% specificity, and only 0.8% error. This robust and scalable architecture offers a high-performance, privacy-aware solution for breast cancer diagnosis in both clinical and remote healthcare settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

IoT-Enabled Breast Cancer Diagnosis Using Visual Geometry Optimized Grounded Non-convolutional Graph Neural Network

  • 期刊论文

  • 发表日期：

    2025年8月5日

查看全文
  • B. Saritha,
  • Dinesh Kumar Anguraj,
  • Elangovan Muniyandy,
  • Krishna Prakash Arunachalam

摘要

Breast cancer remains a leading cause of mortality among women globally, and early, accurate diagnosis is essential for effective treatment. Integrating IoT with intelligent diagnostic systems enables efficient and remote healthcare delivery. Traditional convolution-based diagnostic models often fail to capture the spatial and relational intricacies in mammographic images and lack scalability for real-time use. To address this, we propose a novel IoT-enabled framework leveraging the visual geometry-grounded non-convolutional graph neural network (VG2NCGNN), which incorporates observability-constrained resampling-free cubature Kalman filter for noise-resistant preprocessing, inverse Z-transform, and Wiener–Hopf factorization for advanced feature extraction, and Visual Geometry-Grounded Transformers for spatial reasoning. The model is optimized using the Addax Optimization Algorithm (AOA) to enhance accuracy and convergence. Designed for real-time deployment with IoT-connected mammography scanners, the VG2NCGNN-AOA framework demonstrated superior performance on the BreakHis and MIAS datasets, achieving 99.2% accuracy, 98.6% precision, 98.1% sensitivity, 98.3% F1-score, 98.9% specificity, and only 0.8% error. This robust and scalable architecture offers a high-performance, privacy-aware solution for breast cancer diagnosis in both clinical and remote healthcare settings.