This repository provides pretrained nnU-Net (v2) models for 3D semantic segmentation of vertebral bodies and spinous processes (posterior spinal elements) from binary spine masks (individual vertebra). The models were trained on the Verse20 dataset (cropped to individual vertebrae), which includes annotated cervical, thoracic, and lumbar vertebrae. Segmentations include two labels: 1 = Spinous process 2 = Vertebral body These models are useful for refining vertebral morphology, identifying posterior elements, and enabling downstream applications such as vertebral labeling and finite element modeling. Input: Binary masks of the spine (e.g., merged or coarse spine segmentations).Output: Multi-label segmentations identifying vertebral bodies and spinous processes.