P2G-Spine: A Vertebral Point Cloud Dataset

Read full text Summarise with VIRAT
Authors: 夏潇
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This dataset contains 3D point cloud models of 5526 thoracolumbar vertebrae (T7-L5), derived from the publicly available CTSpine1k CT dataset. Each point cloud is stored as a plain text (TXT) file, with each line representing a single point and containing six floating-point numbers: the first three are spatial coordinates (x, y, z), and the last three are normal vector components (nx, ny, nz). Data generation pipeline: vertebral segmentation from CT (TotalSegmentator) → 3D surface reconstruction → uniform sampling to generate point clouds. All vertebrae have undergone manual quality control to ensure surface integrity and anatomical accuracy. The point clouds are stored in standard formats and can be directly used for pre-training of point cloud deep learning models, geometric feature learning, or self-supervised learning tasks.This dataset is part of the P2G (Point to Guide) project, aiming to provide foundational data resources for research on intelligent navigation in spinal surgery and patient-specific guide design. The dataset contains no patient privacy information and is freely available for academic research.

Fields of Study & Subjects

Medical Imaging and Analysis 3D Shape Modeling and Analysis Anatomy and Medical Technology

Publication details

Journal / source
ScienceDB
Publication date
19 March 2026
Language
EN
Document type
Dataset