GENIUS data

From RealityGrid

MRI Data Sources

There are currently two sources of data we plan to use for the GENIUS project.

1. The IXI Brain Database

From the website www.ixi.org.uk

The IXI database contains brain MR images from 550 normal subjects between the age of 20 and 80 years. The brain MR images acquired include standard anatomical T1 weighted volumes (MP-RAGE) and dual echo (proton density and T2 weighted) fast spin echo images acquired with overlapping slices to produce a densely sampled "pseudo-volume". In addition diffusion tensor MR and a time of flight MR angiograms were acquired for each subject.

Data was collected at three sites on three different scanners. All images are anonymised and have been converted to the NIFTI file format. The image data together with metadata that describes the image properties and includes basic demographic information collected from the subjects (age, gender, handedness, smoking habits etc) is freely available for download subject to a simple licence agreement to ensure appropriate end usage and comply with ethical requirements.

Users can Request Access from the website. A follow up response to an confirmation email enables complete access. As of 3/7/2007 there are 175 records in the database. The database can be searched according the the following criteria:

Cardiac Infarct, Ethnicity, Sex, min height, max hight, min age, max age, Hypertension, min. no. daily cigs., maritial status, diabetes, daily cigs, occupation, ever smoked, right handed, qualification, min weight, max weight. Date of birth is also included.

However, no image data is as yet available. The top of the database page states that:

Please note the database is still under construction. There is no actual data available yet. We intend to be online in a few weeks.

Update 20/7/2007

We have been in touch with Daniel Rueckert from Imperial College London, who is responsible for the database content. He has supplied us with three sets of T1, T2, PD and MRA MRI data. Until the images are placed online, he has kindly offered to supply us with more datasets if need be.

2. National Hospital for Neurology and Neurosurgery (NHNN) Data Acquisition

High resolution datasets are to be obtained by Stefen Brew, the consultant clinician from the NHNN who is a co-investigator on the GENIUS project. The initial dataset used to test HemeLB was obtained on physical media (CD). However, the rapid transfer of MRI data requires both network access (GENIUS Network) along with data anonymisation procedures within the NHNN. We are currently waiting on both these issues concerning access to MRI datasets.

2.1 Data Anonymisation

Any given dataset will contain two components. Private data which is held within the hospital network, and anonymised data which is sent out of the hospital network for flow simulations. This anonymised data will be associated with the private data by a unique key. The data required for flow simulations includes 4 components:

1. A unique identifier,

2. The MR volumetric data,

3. The physical size of this volumetric data,

4. Any initial conditions required for image segmentation, pressure boundary conditions, etc.