A paired dataset of multi-modal MRI at 3 Tesla and 7 Tesla with manual hippocampal subfield segmentations on 7T T2-weighted images

The hippocampus, a region of critical interest within clinical neuroscience, is recognized as a complex structure comprising distinct subfields with unique functional attributes, connectivity patterns, and susceptibilities to disease. Nevertheless, there is a shortfall of available databases featuring manually segmented hippocampal subfield data. While 7 Tesla (7T) MRI yields images with superior anatomical detail compared to the more prevalent 3 Tesla (3T) MRI utilized in clinical practice, its widespread implementation is limited due to prohibitive costs. To address the limited access to high-resolution imaging in the absence of 7T MRI, the ongoing development of algorithms aims to synthesize 7T-like MRI from standard 3T scans.

Here, we disseminate a dataset of whole-brain paired T1-weighted (T1w), T2-weighted (T2w) and resting-state fMRI scans acquired at 3T and 7T scanners, featuring manual hippocampal subfield annotations on the 7T T2-weighted images.

The comprehensive description of the design, acquisition, and preparation of the dataset can be found in "readme.txt" file. Image quality is assessed using quality metrics implemented in MRIQC.

We expect that this dataset will further tackle the challenges of hippocampus segmentation on routine 3T MRI and serve as a valuable resource for research and development in 3T-to-7T, T1-to-T2, and functional MR image synthesis.