Clinica Cloud logbook

[ghisvail]

This discussion aggregates meeting minutes and outcome from unformal exchanges around the Clinica Cloud project.

Clinica Cloud refers to the development effort of providing selected features of Clinica as a web service.

[ghisvail]

5W1H

https://en.wikipedia.org/wiki/Five_Ws

Who

• Clinica team responsible for development
• Clinica community responsible for Q&A and support
• ICM IT responsible for operational maintenance
• Researchers at ARAMIS responsible for early feedback and testing
• Tech-savvy community responsible for later feedback

Where

• Proposal for a proof-of-concept on Inria infrastructure
• Need for a costs-benefits analysis on available deployment strategies (on-premise, hosted by Inria, hosted by ICM, AWS...)
• Which persistence strategy for user supplied data ?
• Need for HDS certification?
• Who has access to the data uploaded ? (c.f VolBrain, NiftyWeb)
• Desirable to have a Clinica Cloud runnable locally. (c.f. MLflow)

When

• Within 2 years.
• Milestones to be determined.
• Maybe benefit from subventions.

What

• Data visualization
• Quality checking
• No-setup run of pipelines
• Help users analyze runs

Why

• Avoid installing dependencies
• Overcome technical knowledge shortcomings
• Expand Clinica's audience towards:
  • Clinicians
  • Users with no access to sufficient resources
• Make results more transparent
• Validate plausibility of results
• Validate completeness of workflows
• Validate the quality of the results
• Simplify (improve) results reporting and viewing

How

• Time efficiency in performing a complete clinical study
• Reduced onboarding time for an internal user
• Percentage of pipelines available in the cloud
• Ease of deployment (local or remote)
• Improved quality of research (constrained user workflow)
• Availability / completeness of user feedback (contextual help)

[ghisvail]

Kick-start Brainstorming

• Identify the use cases and functionalities for Clinica Cloud
• Set their respective priorities in the initial development cycle

Use cases / functionalities

• Dataset sourcing

Interface for uploading or giving access to BIDS datasets which users want to process.

• Pipeline submission

Interface for submitting processing pipelines on BIDS datasets and monitor their progress.

• Results exploration

Interface for browsing, viewing and filtering results produced by processing pipelines.

• Quality assessment

Specialized interface focused on reviewing the quality of processing pipelines.

Priorities

1. Results exploration

2. Quality assessment

3. Pipeline submission

4. Dataset sourcing

[ghisvail]

Meeting with ICM DSI (2022-01-12)

• Communicated the initial set of specifications listed above.
• Clarified the role of the DSI for the ramp-up stage of the project (consultency).
• Since the ICM DSI is invested in AWS, Clinica Cloud should be designed with AWS as a validation target.

[ghisvail]

Meeting with ICM DSI (2022-02-01)

• Presentation of AWS usage within the DSI.
• Investigation of localstack for simulating an AWS cloud locally.
• Requested access to an S3 bucket for initial testing. To be granted and confirmed later.

[ghisvail]

Review of existing UI components

Projects

• niivue
  Example: https://niivue.github.io/niivue-ui/

• Papaya
  Example: https://github.com/rii-mango/Papaya/wiki/User-Interface-Features
  Demo: https://papaya.greenant.net/

• BrainBrowser, AGPL-3.0 licensed, likely unmaintained.
  Example: https://brainbrowser.cbrain.mcgill.ca/#demos

• AMI
  Example: https://fnndsc.github.io/ami/#viewers_upload

Comments

• SHANOIR uses Papaya in their SPA.