Overview

Schema Version

https://atlas.rsna.org/schemas/2025-11/dataset.json

Name

Prostate Imaging: Cancer AI (PI-CAI), version 1.1

Link

https://zenodo.org/record/6667655

Indexing

Keywords: prostate, prostate cancer, clinically significant prostate cancer, MRI, biparametric MRI, T2-weighted imaging, diffusion-weighted imaging, ADC, PI-CAI

Content: GU, MR

RadLex: RID50315, RID49502, RID50307, RID38890, RID50295, RID50313, RID45689, RID50294, RID49531

Author(s)

Saha A

Twilt JJ

Bosma JS

Organization(s)

Radboud University Medical Center

University Medical Center Groningen

Ziekenhuisgroep Twente

Version

1.1

Comments

Annotated public prostate MRI dataset used for model training and internal testing in the study. Consists of 1500 anonymized scans from 1476 patients collected at three centers (Radboud University Medical Center, University Medical Center Groningen, Ziekenhuisgroep Twente).

Date

Published: 2022-06-19

References

[1] Saha A, Twilt JJ, Bosma JS, et al.. "Artificial Intelligence and Radiologists at Prostate Cancer Detection in MRI: The PI-CAI Challenge (Study Protocol). Version 1.1.". . 2022-06-19. Available from: https://zenodo.org/record/6667655

Dataset

Motivation

Training and internal testing for MRI-based csPCa deep learning model development and validation.

Partitioning scheme

Random 4:1 split into training and internal testing, as described in the paper.

Missing information

License type, exact file formats and image resolution not specified in the paper.

External data

External testing used separate, non-public multicenter clinical datasets from three Chinese sites (Cs 1–3); not part of PI-CAI.

Confidentiality

Anonymized public dataset as stated in the paper.

Sensitive data

Medical imaging data; de-identified.