City Research Online

Abstract

The most significant cancer killer across the globe is lung cancer, and the primary cause behind this is that the diagnosis was late and it is not easy to differentiate between malignant and benign lung nodules from the Computed Tomography(CT) scans. Proper and timely detection is highly critical in raising the survival rate of the patients. Still, most of the traditional diagnostic methods are generally expensive and less accurate, and most deep learning methods also lack generalizability, richness of features, and inappropriate handling of imbalanced data. Therefore, this study is intended to create a precise IoT based hybrid deep learning system to detect lung cancer early and accurately from CT scans. A hybrid deep learning IoT-based lung cancer detection system is proposed that involves the ResNet152V2 and InceptionV3 models with transfer learning, focal loss to reduce class imbalance, and a new Hybrid Channel-Spatial Attention (HCSA) module. The use of dual-input approaches by the model enables learning from different augmented views of CT scans and facilitates more effective contextual learning and feature extraction from residual and depthwise separable representations. The system demonstrated here achieved 98% accuracy when tested on the benchmark IQ-OTH/NCCD dataset. This research offers a clinically relevant answer to allow radiologists to diagnose more correctly and quicker, thus decreasing diagnostic errors and patient mortality by early detection of lung cancer.

Publication Type:	Conference or Workshop Item (Paper)
Additional Information:	This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record will be available online at: https://link.springer.com/series/7818
Publisher Keywords:	Lung Cancer, Computed Tomography(CT), Internet of Things (IoT), Hybrid Channel-Spatial Attention (HCSA), Dual-Input Approaches
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science R Medicine > RC Internal medicine > RC0254 Neoplasms. Tumors. Oncology (including Cancer)
Departments:	School of Science & Technology School of Science & Technology > Department of Computer Science
SWORD Depositor:	Symplectic Administrator

Export

Downloads