Labeled Brain Scans
We are building a database of neuroanatomically labeled MRI brain scans. Each year, we’ll label more scans, more and finer-scale regions, and improve the labels we have already created. We’ve created a probabilistic atlas from these scans.
The cost to manually label one scan is $2449. We license for less than that to spread the cost so we can continue to improve and increase our database of results. Nobody pays full price and everybody wins!
If you’d like to see a fully-labeled brain scan before committing to either the academic or commercial data licenses, then download this labeled scan and take a look. The scan will open in a demo version of our in-house manual labeling tool “NVM”: Please click here. If you just want the labeled scan in NIfTI “.nii” format click here.
We offer a unique subscription where you will get many labeled scans each year for the low annual cost of $1500. There are two subscription data sets available:
2012: 63 scans with subject ages from 5 to 96 years old,
2013: 30 ADNI scans, 20 repeat scans (available soon: Colin27 scan)
Summary of individual data sets:
By purchasing an Academic Subscription, you must be from an academic institution and agree to abide by the license (no sharing and no commercial use, see the license file for complete details).
To purchase either of these Academic Subscriptions, please contact us for more information.
If you want to use our data for commercial purposes, such as creating or validating a product or service, then this is the option for you. We have pricing options to accommodate most needs and would be delighted to discuss them with you. Please contact us for more information.
What is the Data and why would you want it?
Our mission is to build and release a “Model” of the human brain as seen in MRI scans. Right now, the main component of the Model is our growing database of manually labeled brains. We also offer a probabilistic atlas. The current 114 scans of subjects ranging in age from 5 to 96 have been painstakingly labeled and double-checked by experts using two protocols that precisely define neuroanatomical region boundaries. The first is based on the “general segmentation” defined by the MGH Center for Morphometric Analysis. The second is a parcellation of the cerebral cortex into regions defined by gyral and sulcal landmarks: brainCOLOR. This is the highest quality and most comprehensive manually labeled brain data available anywhere.
You need this data if you are developing a medical device or image analysis software for the human brain where precise anatomical localization is important such as stereotactic radiosurgery, automated brain labeling algorithms, and neuroanatomical visualization, to name a few application areas. Unlike an atlas of a single subject, this data provides an indication of the variation of the living human brain. A subset of this data was used as the “gold standard” for the MICCAI 2012 Grand Challenge and Workshop on Multi-Atlas Labeling (link).
We presented this poster at the Society for Neuroscience meeting in October 2015 that discusses the reliability of our labeling, and since then we updated our results and the overlap statistics are given here.
Imagine what you could do with our results!