The lung cancer contribution to the collective reason across world for cancer-related death is significantly high even after the development of various early detection methods. Early findings and treatment are crucial and depend on accurate detection of the symptom of lung lumps in CT scan images. In this work, a new architecture for large lung nodule detection is presented by developing a convent architecture designed specifically for its maximization lung nodule detection accuracy. To provide optimal performance for this particular medical imaging task, the method proposed employs evolutionary methods to optimize the CNN architecture and hyper parameters. Enhancing prognosis and survival rates for lung cancer patients requires early detection of lung nodules in computed tomography (CT) scans. Accomplishing extraordinary accuracy and low false-positive rates is difficult for traditional methods. To increase the correctness and reliability of lung nodule detection, we propose a new approach that combines convolutional neural networks (CNN) and grey wolf optimisation (GWO). Stringent tests on CT scan datasets validate the excellence of the GWO-CNN approach. The optimised CNN works far better than the traditional approaches, as can be seen from the results, which indicate improvement in the minimization of false positives, false negatives and enhanced detection accuracy. This hybrid approach not only represents a significant leap forward in computerized lung cancer screening technology, but also has the potential to integrate deep learning with meta-heuristic optimisation. Lastly, the architecture of GWO-CNN carries with it prospects of improved earlier lung cancer detection and bettering clinic results through presenting a coherent and accurate avenue to detecting lung nodules.