Skip to content

T2-MRI Tutorial

Cloud HyperASHS Tutorial: T2-MRI

This tutorial uses the HyperASHS-PMC service. Make sure that your MRI data is compatible with this model. The procedure is the same for the HyperASHS-HSG-Body service, which follows the harmonized hippocampal body protocol of the Hippocampal Subfields Group.

ITK-SNAP 4.6 or later is required

HyperASHS returns segmentations at a finer resolution than the input MRI, i.e., in a different voxel space from the main image. Only ITK-SNAP 4.6 and later can work with such segmentations, so older versions of ITK-SNAP cannot be used with Cloud HyperASHS.

1. Create an ITK-SNAP workspace

To send your images to the cloud, you need to create an ITK-SNAP workspace that contains the T2-weighted and T1-weighted images.

  • Open ITK-SNAP
  • Use File->Open Main Image... to load the T2-weighted image as the main image
  • Use File->Add Another Image... to load the T1-weighted image as the secondary image
  • Save the ITK-SNAP workspace to a file using Workspace->Save Workspace

HyperASHS registers the T1w image to the T2w image automatically, so you don't need to align them first.

ITK-SNAP with T2 and T1 images loaded

ITK-SNAP with a T2-weighted image loaded as the main image and the T1-weighted image loaded as the additional image.

2. Log in to the Distributed Segmentation Service (DSS)

Next, you need to log in to the DSS server. This requires a Google account.

  • Open the DSS window using Tools->Distributed Segmentation Service
  • Select the "Connect" tab if it is not already selected
  • Press the Get Token button. This opens a web browser at the DSS login page, shown below.
    • Follow the login prompts on the website. Once you log in with your Google account, you will see a 40-character login token. Copy the token and paste it back into ITK-SNAP.
  • The ITK-SNAP DSS window should show an updated status, in green, indicating that you have logged in
    • The next time you open ITK-SNAP on the same device, your login information will be remembered.

DSS window, connect tab

The ITK-SNAP DSS window, configuration tab. Press "Get Token" to log in.

DSS login web page

The login web page. Paste the token into ITK-SNAP.

3. Upload your workspace and create a ticket

In this step, you send your workspace to the cloud for segmentation with HyperASHS.

  • Select the "Submit" tab in the ITK-SNAP DSS window
  • Select the HyperASHS-PMC service from the list (or HyperASHS-HSG-Body)
  • In the table of "tags", make sure that the correct images are assigned to the "T2-MRI" and "T1-MRI" tags
  • Press Submit. The workspace will be uploaded to the server and a new ticket will be created.

4. Monitor ticket status

Once a ticket is created, the DSS window switches to the "Results" tab, where you can monitor the progress of your ticket. While the ticket is being processed, you can close ITK-SNAP or submit additional tickets.

  • Your ticket is highlighted in the list of tickets on the left of the window, and its progress is shown on the right.
  • Note the status of your ticket:
    • ready: the ticket has been uploaded and is in the queue to be processed
    • claimed: the ticket is being processed
    • success: the ticket was processed successfully and results are available
    • failed/timed out: an error occurred while servicing the ticket
  • Once your ticket is in the claimed state, log messages show the progress of the segmentation. Some log messages have attachments: the registration and segmentation QC images. Click the attachment button to list the attachments and open them in a web browser.
  • To abort ticket processing, press the Delete button

5. Download and examine the results

Once a ticket completes successfully, you can download the results as an ITK-SNAP workspace.

  • Select a ticket that is in the success state and press the Download button
  • Choose where to save the result workspace and whether to open it after downloading
  • Use the Layer Inspector in ITK-SNAP (Tools->Layer Inspector) to see the segmentation layers in the workspace

See Outputs for a description of the HyperASHS results.

Using classic ASHS? See the equivalent page for ASHS Classic.