Integrating SURF Feature Reduction with CCapsNets Classification for improving the accuracy of Lung Cancer Detection

Abstract

Lung cancer continues one of the indicating causes of cancer-linked transience worldwide, compelling the advancement of accurate and efficient investigative tools. In this research, we propose a novel approach for lung cancer detection utilizing feature reduction with Speeded-Up Robust Features (SURF) and classification with Classification Capsule Networks (CCapsNets). The researcher conduct experiments on the UC Irvine Machine Learning Repository lung cancer dataset, which comprises a diverse collection of computed tomography (CT) images. Firstly, SURF is used to isolate robust and discriminative attributes from the lung CT images. SURF's ability to detect local features invariant to scale and rotation enables effective representation of the lung tissue characteristics. Next, CCapsNets us utilized, a state-of-the-art deep learning architecture known for its ability to capture hierarchical relationships within data, for lung cancer classification. CCapsNets leverage capsule networks to preserve spatial hierarchies and improve generalization performance, particularly in medical image analysis tasks. The investigational results exhibit the efficacy of the intended methodology in lung cancer detection. By integrating SURF feature reduction with CCapsNets classification, superior accuracy og 98.6% is achieved in evaluation to traditional methods. Furthermore, the interpretability of CCapsNets enables insights into the learned features and contributes to the understanding of lung cancer imaging biomarkers. This research work presents a promising framework for lung cancer detection, leveraging advanced image processing techniques and deep learning methodologies. The proposed approach holds significant potential for enhancing early diagnosis and prognosis prediction in clinical settings, thereby improving patient role conclusions and reducing the liability of lung cancer morbidity and mortality.