The three‐dimensional (3D) anatomy of Kambin's triangle is crucial for surgical planningin minimally invasive spine surgery via the transforaminal approach. Few pieces ofresearch have, however, used image segmentation to explore the 3D reconstructionof Kambin's triangle. This study aimed to develop a new method of 3D reconstruction ofKambin's triangle based on automated magnetic resonance image (MRI) segmentationof the lumbar spinal structures. An experienced (>5 years)“ground truth”spinal painphysician meticulously segmented and labeled spinal structures (e.g., bones, dura mater,discs, and nerve roots) on MRI. Subsequently, a 3D U‐Net algorithm was developed forautomatically segmenting lumbar spinal structures for the 3D reconstruction ofKambin's triangle. The Dice similarity coefficient (DSC), precision, recall, and the areaof Kambin's triangle were used to assessanatomical performance. The automaticsegmentation of all spinal structures at the L4/L5 levels and L5/S1 levels resulted ingood performance: DSC = 0.878/0.883, precision = 0.889/0.890,recall = 0.873/0.882.Furthermore, the area measurements of Kambin's triangle revealed no significantdifference between ground truth and automatic segmentation (p= 0.333 at the L4/L5level,p= 0.302 at the L5/S1 level). The 3D U‐Net model used in this study performedwell in terms of simultaneous segmentation of multi‐class spinal structures (includingbones, dura mater, discs, and nerve roots) on MRI, allowing for accurate 3Dreconstruction of Kambin's triangle.