The lung exhibits a complex hierarchical structure comprising lobes and pulmonary segments. With recent advancements in lung cancer treatment, there’s a growing demand for precise segmentation between lung segments. However, there is a challenge that segments are not differentiated by visual features, but outlined by invisible borders constructed from the border of each lobe, bronchus, pulmonary artery, and vein. To tackle the issue, we introduce a novel framework for determining intersegmental border within a lobe, the Pulmonary Segment segmentation model based on the point-cloud Gaussian Mixture Model (PSGMM). PSGMM takes the bronchus, artery, vein, and lobe surface in point-cloud form, to construct the probability map of each segment in the form of a Gaussian mixture model. PSGMM is designed to emulate a physician’s examination mechanism by considering anatomical features and provides reliable and promising results.