metadata

license: cc-by-nd-4.0
size_categories:
  - n>1T
tags:
  - medical
dataset_info:
  features:
    - name: ICT
      dtype: image
    - name: LDCT_Low
      dtype: image
    - name: LDCT_Mid
      dtype: image
    - name: LDCT_High
      dtype: image
    - name: LACT_Low
      dtype: image
    - name: LACT_Mid
      dtype: image
    - name: LACT_High
      dtype: image
    - name: SVCT_Low
      dtype: image
    - name: SVCT_Mid
      dtype: image
    - name: SVCT_High
      dtype: image
  splits:
    - name: train_previews
      num_bytes: 62199112
      num_examples: 44
    - name: test_previews
      num_bytes: 16108271
      num_examples: 11
  download_size: 153191938
  dataset_size: 78307383
configs:
  - config_name: default
    data_files:
      - split: train_previews
        path: data/train_previews-*
      - split: test_previews
        path: data/test_previews-*

SimNICT

SimNICT is the first dataset for training universal non-ideal measurement CT (NICT) enhancement models.
The dataset comprises over 10.9 million NICT-ICT image pairs, including low dose CT (LDCT), sparse view CT (SVCT), and limited angle CT (LACT), under varying defect degrees across whole-body regions.
We have currently uploaded part of the SimNICT dataset, SimNICT-AMOS-Sample, with preview images in the dataset viewer. The complete SimNICT dataset will be gradually uploaded in future releases.

SimNICT-AMOS-Sample

SimNICT-AMOS-Sample dataset contains 55 ICT volumes from the AMOS dataset in SimNICT, and each ICT volume has been simulated using the same NICT simulation method as in SimNICT, generating 9 types of NICT volumes.
This dataset is divided into training and test sets, with 20% and 80% of the total volumes, respectively, to evaluate the performance of our proposed TAMP model.

Source Dataset Statistics

SimNICT starts from the ICT images from ten publicly CT datasets that encompass whole-body regions.
By removing low-quality volumes, our SimNICT dataset finally obtains 3,633,465 images from 9,639 ICT volumes.

Source	Provenance	Volume	Slice	License
COVID-19-NY-SBU	TCIA	459	118,119	CC BY 4.0
STOIC	Grand Challenge	2,000	867,376	CC BY 4.0
MELA	Grand Challenge	1,100	496,673	CC BY 4.0
LUNA	Grand Challenge	888	227,225	CC BY 4.0
LNDb	Grand Challenge	294	94,153	CC BY 4.0
HECKTOR22	Grand Challenge	883	200,100
CT_COLONOGRAPHY	TCIA	1,730	938,082	CC BY 3.0
AutoPET	Grand Challenge	1,014	560,796
AMOS	Grand Challenge	500	76,679	CC BY 4.0
CT Images in COVID-19	TCIA	771	54,262	CC BY 4.0

Ongoing

Release the SimNICT dataset containing 10.9 million NICT-ICT image pairs.
Release the SimNICT-AMOS-Sample dataset, a subset of the SimNICT dataset.

Citation

@misc{liu2024imagingfoundationmodeluniversal,
      title={Imaging foundation model for universal enhancement of non-ideal measurement CT}, 
      author={Yuxin Liu and Rongjun Ge and Yuting He and Zhan Wu and Chenyu You and Shuo Li and Yang Chen},
      year={2024},
      eprint={2410.01591},
      archivePrefix={arXiv},
      url={https://arxiv.org/abs/2410.01591}, 
}