PPCNet is a research dataset containing lumbar spine CT volumes, segmentation masks, biplanar digitally reconstructed radiographs (DRRs), projection matrices, and ground-truth 3D point clouds. Access is provided for academic, scientific, educational, and non-commercial research purposes, subject to the PPCNet Dataset License and applicable upstream dataset terms. Please provide the information below. Access requests are manually reviewed by the PPCNet dataset authors.

By requesting access, you acknowledge that you have read and agree to the PPCNet Dataset License and applicable upstream dataset terms.

PPCNet Dataset

Biplanar DRRs + Projection Matrices + Ground-Truth Point Clouds

A curated lumbar spine dataset for 3D point cloud reconstruction from biplanar radiographs.

Overview

This dataset provides paired biplanar DRRs, calibrated 3×4 projection matrices, and dense ground-truth point clouds for 1,037 patients with complete L1–L5 lumbar vertebrae, derived from the publicly available VerSe'19, VerSe'20, and CTSpine1K collections.

Dataset Access

PPCNet is publicly hosted on the Hugging Face Hub and is available for academic, scientific, educational, and non-commercial research purposes.

Access: Access to the PPCNet dataset files is provided through a gated, manually reviewed access process. Researchers may request access through the Hugging Face repository by providing the requested information and agreeing to the PPCNet Dataset License and applicable upstream dataset terms.

This access-control mechanism does not change the dataset's documentation, citation, or public repository availability. It applies to access to the dataset files.

Request access: PPCNet Dataset on Hugging Face

📁 Dataset Structure

Each of the 1,037 patient folders contains:

File	Description
`ct.nii.gz`	CT volume (LPS orientation)
`seg.nii.gz`	Segmentation labels (L1=20, L2=21, L3=22, L4=23, L5=24)
`gt_ppc.vtk`	Ground-truth point cloud (5,120 points, world-mm)
`AP_0/drr_AP_0.png`	Anteroposterior DRR (512×512)
`AP_0/P_AP_0.txt`	3×4 AP projection matrix
`LP_90/drr_LP_90.png`	Lateral DRR (512×512)
`LP_90/P_LP_90.txt`	3×4 lateral projection matrix

Label mapping: L1=20, L2=21, L3=22, L4=23, L5=24

Data Split

Fixed split in dataset_split.json (seed=42):

Split	Patients	Usage
Train	829	Model training
Validation	103	Hyperparameter tuning
Test	105	Final evaluation
Total	1,037

DRR Generation

DRRs are generated using Plastimatch's ray-casting algorithm with the following parameters:

Parameter	Value
Source-to-Axis Distance (SAD)	1,000 mm
Source-to-Imager Distance (SID)	1,500 mm
Detector Size	500 × 500 mm
Image Resolution	512 × 512 px
Views	AP (0°) + Lateral (90°)
Bone Enhancement	2.5× for HU > 300
Post-processing	CLAHE normalisation

Each DRR comes with a calibrated 3×4 perspective projection matrix encoding the exact source–detector geometry.

Ground-Truth Generation

The ground-truth point clouds are constructed through a five-step pipeline:

Surface extraction — marching cubes on per-vertebra segmentation masks
Coordinate transform — vertices mapped to physical (mm) space via NIfTI affine
Concatenation — all five vertebrae (L1–L5) merged into a single cloud
Alignment — rigid ICP with four flip initialisations to match DRR projection space
Sampling — farthest-point sampling to a uniform 5,120-point representation

⬇️ Download

from huggingface_hub import login, snapshot_download

# Authenticate with your Hugging Face account.
# Your account must have approved access to PPCNet.
login()

snapshot_download(
    repo_id="ppcnet-dataset/PPCNet",
    repo_type="dataset",
    local_dir="./PPCNet_Dataset"
)

Or use the Hugging Face CLI:

pip install huggingface_hub
huggingface-cli download ppcnet-dataset/PPCNet --repo-type dataset --local-dir ./PPCNet_Dataset

Source Datasets

This dataset builds upon:

📦 VerSe — A Vertebrae Labelling and Segmentation Benchmark for Multi-Detector CT Images
📦 CTSpine1K — A Large-Scale Dataset for Spinal Vertebrae Segmentation in CT

Citation

If you use this dataset in your work, please cite:

@misc{ppcnet_dataset_2026,
    author       = { PPCNet Dataset },
    title        = { PPCNet (Revision bce5fd9) },
    year         = 2026,
    url          = { https://huggingface.co/datasets/ppcnet-dataset/PPCNet },
    doi          = { 10.57967/hf/9944 },
    publisher    = { Hugging Face }
}

License and Terms of Use

PPCNet is distributed under the PPCNet Dataset License and Terms of Use.

The dataset is provided for academic, scientific, educational, and non-commercial research purposes, subject to the terms of the PPCNet Dataset License and the applicable terms of the underlying source datasets.

Users are responsible for reviewing and complying with the applicable licenses and terms of the upstream datasets, including VerSe'19, VerSe'20, and CTSpine1K.

Downloads last month: 821

Total file size:

69.2 GB