Pulpy3D Dataset
This page describes the dataset Pulpy 3D associated with the pulp segmentation benchmarking efforts made by Knights Lab and is part of MICCAI2024. The dataset described here is an extension of the previously released dataset ToothFairy.
Features
The table below includes core information about the Pulpy 3D dataset released for MICCAI2024. The dataset provides semantic and instance annotations for the pulp. The dataset used the original annotations for the IAN from ToothFairy dataset.
| Field | Value |
|---|---|
| File Format | NIFTI (.nii.gz) |
| Dataset Size | 443 Scans |
| Values Scale | Hounsfield |
| Dataset Classes | 19 (the complete list is available in the dataset.json file) |
| Avg Volume Shape | (169, 342, 370) |
| Max Volume Shape | (178, 423, 463) |
| Min Volume Shape | (148, 265, 312) |
Dataset Structure
The Pulpy3D dataset consists of three components: raw images, corresponding segmentation maps, and a dataset.json file specifying some metadata. The class IDs are an extension of the FDI World Dental Federation notation and include: Background, Inferior Alveolar Canals, Lower Teeth (Wisdom Teeth included), and Anomaly Teeth, for a total of 19 classes.
