Neurosense is a trained neural network for identifying full or near-full coverage head/brain scans in large human magnetic resonance image datasets.

Quick Start Guide

Preamble

Neurosense comes in the form of an Ubuntu Virtual Machine, which has FSL pre-installed. Therefore, its use is conditional on your agreement of the FSL software license available at https://fsl.fmrib.ox.ac.uk/fsldownloads_registration. You do not need to download FSL itself as it has been pre-installed. However, make sure you read and agree to the terms of the FSL software license before using Neurosense.

Requirements

You will need VirtualBox to run the Neurosense virtual machine. If you do not already have VirtualBox, download and install it for your platform from https://www.virtualbox.org/wiki/Downloads

Installation and Use

• Download Neurosense_v0.3.ova from this link (password: neurosense) and save it in an appropriate location. It is recommended that all work is carried out in an encrypted, safe environment if you will be using Neurosense on sensitive data.

• Start VirtualBox, and choose File/Import Appliance to import Neurosense.ova

• Once imported, right click on Neurosense in VirtualBox and choose Settings. ​

• Within Neurosense – Settings click on Shared Folders and click the small add icon on the right.

• Within the Add Share dialog, choose the folder where the NIFTI volumes you would like to run through Neurosense are located.

• Put myfolder for Folder Name and check Auto-mount.

• Start the virtual machine and login as user neurosense (password: neurosense)

• Type the following commands to make sure you can see your files/folders: ls /media/sf_myfolder ​

• Type the following commands to run Neurosense on your files/folders: neurosense /media/sf_myfolder ​

Notes

• You can add --pdf as a parameter if you would like Neurosense to produce a Pdf report.
• You may wish to check your VirtualBox settings to ensure you have enough computing resources available to Neurosense.
• Neurosense has been trained primarily using T1-weighted MRI volumes, performance on MR volumes of other contrasts and non-MRI volumes will be degraded.
• Any queries can be sent to [email protected].

Sample Output

Acknowledgements

If you use Neurosense for your research, please acknowledge our publication below:

Kanber, B., Ruffle, J., Cardoso, J. et al. Neurosense: deep sensing of full or near-full coverage head/brain scans in human magnetic resonance imaging. Neuroinform (2019) doi:10.1007/s12021-019-09442-x

Notices

Commercial use is not allowed without explicit permission.