Lung cancer remains one of the leading causes of cancer-related mortality worldwide, requiring precise and early detection methods. A hybrid deep learning model is presented for detecting and classifying adenocarcinoma, a common subtype of NSCLC. Initially, a pre-trained model for lung nodule detection is developed using the LIDC-IDRI dataset, effectively distinguishing between benign and malignant nodules. The NSCLC Radiogenomic dataset is utilized to classify adenocarcinoma. Lung segmentation is performed using a watershed algorithm and Hounsfield unit (HU)-based thresholding to extract lung regions from 3D CT images. The pre-trained model detects malignant nodules, and their features are subsequently analyzed for classification as adenocarcinoma or other NSCLC types. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) is employed during the classification phase. The proposed method provides a efficient framework for the detection and classification of adenocarcinoma, improving accuracy and supporting clinical decision-making.