This research investigates whether active learning can reduce the amount of data that must be labelled for lung segmentation. It is based on the challenge that machine learning models require large amounts of annotated CT image data, while manual annotation is slow, expensive, and often requires expert input. To address this, the study will apply active learning to a machine learning model that performs lung segmentation, with the aim of selecting only the most informative samples for annotation and determining whether results close to full dataset annotation can still be achieved.