Calgary-Campinas Public Brain MR Dataset
A collaborative effort between researchers at the Vascular Imaging Lab located at the University of Calgary and the Medical Image Computing Lab located at the University of Campinas (UNICAMP) originated the Calgary Campinas public brain magnetic resonance (MR) images dataset. The dataset was first released with the following publication:
- Roberto Souza, Oeslle Lucena, Julia Garrafa, David Gobbi, Marina Saluzzi, Simone Appenzeller, Letícia Rittner, Richard Frayne, and Roberto Lotufo. "An open, multi-vendor, multi-field-strength brain MR dataset and analysis of publicly available skull stripping methods agreement."NeuroImage 170 (2018): 482-494.
If you use this dataset in your experiments, we ask you to kindly cite in your research the above mentioned paper.
- Congratulations to the team ResoNNance composed by Jonas Teuwen, Matthan Caan, Dimitrios Karkalousos and Nikita Moriakov. They were the winners of both tracks of our MC-MRRec challenge! The team members are affiliated with the Netherlands Cancer Institute, the Radboud University Medical Center, and the Amsterdam UMC . Thank you for everyone who participated of the challenge. We had a great session and literally closed the 2020 MIDL conference. Please check our new blog post.
- We just had a paper investigating longitudinal data integration during MR acquisition and reconstruction in the IEEE Journal of Selected Topics in Signal Processing (JSTSP). The paper should be available online shortly.
- Roberto Souza, Youssef Beauferris, Wallace Loos, R. Marc Lebel and Richard Frayne. "Enhanced Deep-learning-based Magnetic Resonance Image Reconstruction by Leveraging Prior Subject-specific Brain Imaging: Proof-of-concept using a Cohort of Presumed Normal Subjects", IEEE JSTSP, 2020 (ACCEPTED)
- Our dataset is now available for download both from OneDrive and GDrive. The raw MR data are also available from Amazon Simple Storage Service. If your old download links expired, just submit the download form again.
The goal of this dataset is that the scientific community use it to develop innovative and fast big data (i.e. deep learning) models to reconstruct, process and analyse brain magnetic resonance (MR) images. The key for developing big data applications is to have good data and representative of the variability we encounter in real applications. In the initial Calgary-Campinas (CC) dataset release we provided T1 volumes acquired in 359 subjects on scanners from three different vendors (GE, Philips, and Siemens) and at two magnetic field strengths (1.5 T and 3 T). The scans correspond to older adult subjects. The second dataset release included raw MR data.
Medical imaging data is continually revisited to guide decisions around future episodes of care. That is currently not the standard for image acquisition and reconstruction. We need better data integration strategies to optimize image acquisition and reconstruction. We are working on that. We hypothesize that we can leverage past subject-specific imaging information to enhance the reconstruction of following imaging exams (see figure below). We start our investigation with MR, but there is no reason that this paradigm cannot be applied across modalities. The raw MR data we currently provide corresponds to a single snapshot in time of a subject's brain. Our first edition of our online MR reconstruction challenge focus on reconstructing that brain snapshot from undersampled data with high fidelity compared to the fully sampled reconstruction. We are acquiring longitudinal data that we expect to make publicly available shortly. In the second edition of the challenge, we want to see if our hypothesis is confirmed (i.e. leverage longitudinal information to enhance reconstruction).
Path defined by the red arrows are triggered only if previous MRI scan available. PACS = Picture archiving and communication system.
Updated: 10 June 2020