Lung cancer is the most diagnosed cancer after breast cancer. Computed Tomography (C.T.) scan is the most used imaging technique in the medical field to diagnosis lung cancer. Still, it is difficult to infer and not identify benign and malignant nodules based on imaging tests. In such cases, doctors advise or recommend lung biopsy. Manual pathological detection systems are a time-consuming, tedious, exhausting task and may lead to medical error; hence computer-aided diagnosis is preferred to obtain better results. In this work, we have collected data of 97 patient records that did a lung biopsy. First, images are prepared using histopathology. Next, stain normalization, image processing techniques, and data augmentation are applied to prepare the dataset. This research work demonstrates an approach to designing and training various variants of the Convolutional neural network to diagnose Lung Cancer in histopathological images.