Segmentation is one of the fundamental tasks in biomedical image processing. Adequate image segmentation and computer-aided diagnosis systems are excellent allies for healthcare professionals. There are multiple methods for image segmentation using image processing techniques that are still being used and developed. These have advantages over machine learning models and deliver reliable and fast results as training data for their operation do not limit them. This work proposes a 3-step semi-automatic pipeline for lung computed tomography image segmentation. It starts with preprocessing, in which the input image is enhanced; then, the image is segmented using the region growing technique, and finally, the segmentation mask is enhanced by applying a hole-filling process. The experimental results of the pipeline provided a Dice Coefficient of 0.9633 and an Intersection over Union of 0.9341 on average.