Comparative Analysis of Lung Cancer Detection Using CNN, KNN, and SVM Algorithms: Evaluating Accuracy and Performance

Lung cancer remains a leading cause of mortality, emphasizing the crucial need for early detection to improve survival rates. Computed tomography (CT) imaging stands out as a pivotal strategy in this endeavor. However, the escalating volume of CT images necessitates efficient and accurate automated detection methods to mitigate manual labeling efforts and ensure consistent results. This research paper employs a comprehensive strategy, merging traditional machine learning models (including KNN and SVM) with state-of-the-art feature extraction methods such as PCA, HOG, and VGG16 transfer learning, as well as utilizing a dedicated Convolutional Neural Network (CNN) for classification. By amalgamating these methodologies, the study seeks to comprehensively evaluate their efficacy in accurately diagnosing lung cancer. The anticipated outcomes aim to contribute novel insights to medical image analysis, particularly in advancing early cancer detection. Through this research, valuable contributions are expected to be made towards the enhancement of medical image analysis techniques, thereby pushing the boundaries of early cancer detection and ultimately improving patient outcomes. The best result of the proposed model shows the accuracy percentage of the SVM classifier varies from ${9 0 \%}$ to ${9 6 \%}$, the accuracy of the KNN classifier varies from 95% to 100%, and the accuracy of the CNN classifier varies from ${9 6 \%}$ to ${9 9 \%}$.