Lung cancer remains the leading cause of cancer-related deaths worldwide. This is mainly because lung nodules, the typical signs of the disease, are hard to detect at an early stage. These nodules are often subtle and have different shapes, making it very challenging to classify them correctly by the use of conventional radiological techniques. Besides, manual interpretation of high-resolution CT scans is a very long process and the results are dependent on the different radiologists’ opinions, which can delay the diagnosis and treatment of the disease. To overcome these problems, less reliant on radiologists and to quicken the diagnosis procedure, this work presents a novel Hybrid Deep Learning Framework that effectively combines the hierarchical feature extraction of the VGG-16 convolutional neural network and the spatial awareness and pose-preserving properties of Capsule Networks. The proposed framework is built to improve the classification of lung nodules into malignant or benign by spatially preserving the hierarchies and making reasonable the image distortions caused by rotation and tiling—problems for which traditional CNN-based models cannot usually give a solution. Proposed framework architecture is constructed in such a way that dual-path strategy is implemented where deep features are extracted with the help of VGG-16, after which Capsule Networks layers receive the data, encode spatial relations, and execute dynamic routing for classification. Such a framework is both trained and tested on the publicly accessible LIDC-IDRI dataset comprising the CT scans of lung nodules that are annotated. To make the model robust and generalizable, the researchers conduct thorough preprocessing that includes converting grayscale images to RGB, rescaling, and data augmentation. These achievements are far beyond those of the existing frameworks and thus, they are strong evidence of the capacity of Proposed framework as a clinically deployable tool for the early screening and diagnosis of lung cancer with high precision.