TIPs:

An accurate and automated AI tool called TIPs for Tooth Instances and Pulp segmentation from CBCT.

Introduction

TIPs works out-of-the-box without requiring any retraining. By inputting a CBCT image, users can obtain both semantic and instance segmentation for teeth and pulps. The final instance labeling follows the FDI World Dental Federation notation.

Installation

Requirements: Ubuntu 20.04, CUDA 11.8

1. Create a virtual environment: conda create -n tips python=3.10 -y and conda activate tips
2. Install Pytorch 2.0.1: pip install torch==2.0.1 torchvision==0.15.2 --index-url https://download.pytorch.org/whl/cu118
3. Install Mamba: pip install causal-conv1d>=1.2.0 and pip install mamba-ssm --no-cache-dir
4. Download code: git clone https://github.com/TaoZhong11/TIPs
5. cd TIPs and run pip install -e .

sanity test: Enter python command-line interface and run

import torch
import mamba_ssm

Model download

https://drive.google.com/file/d/1UuFgZ-kwRryPC-vK7w64xX0VO4iOAeGt/view?usp=drive_link

Model path setting

vi ~/.bashrc
export nnUNet_raw = "/home/path/to/TIPs/nnUNet_raw"  # raw_data_path
export nnUNet_preprocessed = "/home/path/to/TIPs/nnUNet_preprocessed"  # preprocessed_data_path
export nnUNet_results = "/home/path/to/TIPs/nnResults"  # models_path
source ~/.bashrc

Inference

python   TIPs.py   folder_to_be_processed

Training

The training code and data will be made available following the acceptance of the paper. Thank you!

Paper

@article{TIPs,
    title={TIPs: Tooth Instances and Pulp segmentation based on hierarchical extraction and fusion of anatomical priors from cone-beam CT},
    author={Tao Zhong, Yang Ning, Xueyang Wu, Li Ye, Chichi Li, Yu Du, and Yu Zhang},
    journal={Under Review},
    year={2024}
}

Docker

Coming soon.

Acknowledgments

We acknowledge all the authors of the employed public datasets, allowing the community to use these valuable resources for research purposes. We also thank the authors of nnU-Net, U-Mamba and Mamba for making their valuable code publicly available.