Lung cancer is still among the top deadliest diseases across the globe. Early detection constitutes the primary means to survivability. The paper presents an improved Computer-Aided Diagnosis (CAD) system that detects lung cancer at an early stage based on chest X-ray images by utilizing an advanced technique of machine learning in diagnosis. The proposed system works in six main processes: image acquisition, enhancement, segmentation, feature extraction, feature selection, and classification. It works well by utilizing the classifiers Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA); hence, it achieves 100% accuracy in terms of normal-abnormal region discrimination and 94.1% effectiveness in distinguishing malignant benign nodules. The major contribution of the proposed work is the optimized feature selection process, which significantly diminishes computation complexity while maintaining precision in diagnosis. It thus indicates an opportunity with probable impact toward CAD as sufficiently reliable second-opinion tools for radiologists in resource-lean settings. We propose the incorporation of this CAD system into the clinical workflow with a view to availing early diagnosis with reduced chances of human error and ultimately enhancing patient outcomes.