Lung cancer stands as one of the most formidable tumor types, demanding a meticulous and time-intensive research process for detection and categorization. The complexity is exacerbated by the diverse characteristics of lung nodules and their visual similarity to surrounding tissues. Traditional machine learning-based artificial intelligence methods for medical imaging analysis applications often address these challenges in isolation or rely heavily on human interpretation, potentially overlooking subtle relationships among features. By harnessing the hierarchical structures of deep learning-based artificial intelligence techniques, this framework introduces an effective approach for classifying pulmonary nodules in Computerized Tomography (CT) images. The methodology commences with comprehensive data pre-processing techniques to ensure effectual training data. Subsequently, T-Net is employed for lung nodule detection and segmentation. Following this, a CenterNet-based approach extracts intensity and textural features from the segmented images. Subsequently, a NasNet-based classification module assigns labels to nodules as cancerous or non-cancerous based on the extracted attributes. The performance of the method is then assessed on the Lung Image Database Consortium dataset using various evaluation metrics. For the Segmentation task, it achieved a sensitivity of 97.93, a positive predictive value of 96.15, and a dice similarity coefficient of 97.89. For the Cancer Classification task, it attained an F1-score of 99.01, a precision of 99.01, a specificity of 99.04, a sensitivity of 99.10, and an accuracy of 99.13.