Cloud HyperASHS: Overview¶
The easiest way to try out HyperASHS is through our web-based Distributed Segmentation Service (DSS), the same system used by Cloud ASHS. You don't need a GPU or any software besides ITK-SNAP. Your images are anonymized, encrypted and sent to a secure server, where they are processed at the Penn Image Computing and Science Laboratory (PICSL) at the University of Pennsylvania. This service is currently provided for free.
Available services¶
| DSS service | HyperASHS model | Input MRI |
|---|---|---|
| HyperASHS-PMC | PMC_Atlas | T2w + T1w MRI |
| HyperASHS-PMC-T1 | PMC_T1_Atlas | T1w MRI only |
| HyperASHS-HSG-Body | HSG_Body_Atlas (HSG harmonized protocol) | T2w + T1w MRI |
Requirements¶
- You will need to install ITK-SNAP 4.6 or later. HyperASHS returns segmentations in a different (finer) voxel space than the T2-weighted MRI, and earlier versions of ITK-SNAP cannot handle this
- You need a Google account (e.g., Gmail) to log in to the server
- Your data must be compatible with the model behind the service you choose (see MRI Data)
GUI or command line?¶
The easiest way to run HyperASHS in the cloud is with the graphical user interface (GUI) in ITK-SNAP. With just a few clicks you can send your images to the server for processing. Follow one of the tutorials below, depending on what kind of MRI data you want to segment:
- Cloud HyperASHS GUI tutorial: hippocampal subfields from T2-weighted and T1-weighted MRI
- Cloud HyperASHS GUI tutorial: medial temporal lobe subregions from T1-weighted MRI only
If you need to process several datasets from a script, use the command-line tool itksnap-wt, which is installed as part of ITK-SNAP:
For large projects, please install HyperASHS locally. Our server is set up to handle only a moderate number of requests.
Using classic ASHS? See the equivalent page for ASHS Classic.