|
| 1 | +# Clairvoy Tutorial |
| 2 | + |
| 3 | +## Welcome to Clairvoy. |
| 4 | + |
| 5 | +Clairvoy is an experimental [brainlife.io](https://brainlife.io) App. |
| 6 | +The Clairvoy App utilizes a machine learning, prognostic model to predict whether someone who has experienced a concussion is at risk for slow recovery (defined as not cleared for full return to sport at greater than 28 days). |
| 7 | +The diffusion-weighted and T1-weighted MRI should be collected 24-48 hours after a diagnosed concussion. |
| 8 | +This experimental tool is entirely based on the publication of Bertó et al. 2024 (Neuroimage: Clinical, In Press). |
| 9 | +Clairvoy was funded by the U.S. Department of Defense, award W81XWH-20-1-0717, PI: Nicholas L Port. |
| 10 | +The dataset analyzed by Bertó et al. is from the NCAA/DOD CARE Consortium and was jointly funded by the NCAA and DOD, award W81XWH-14-2-0151, PI: Thomas W McAllister. |
| 11 | + |
| 12 | +We encourage clinicians and scientists to try Clairvoy and appreciate any feedback on how to refine the app. |
| 13 | +Drs. [Port ](mailto:[email protected]) and [Pestilli ](mailto:[email protected]) are actively seeking funding to conduct a clinical trial to validate Clairvoy, but, at present, Clairvoy has not been validated for clinical use. |
| 14 | +Thank you for trying Clairvoy and helping us refine the application. |
| 15 | + |
| 16 | +## Directions |
| 17 | + |
| 18 | +The Clairvoy App lives inside the [brainlife.io](https://brainlife.io) platform, an easy-to-use, web-based, Open Science data analysis platform focused on neuroimaging. |
| 19 | +Free and open to anyone in the world, Brainlife has over 1,000 users across dozens of countries. Brainlife offers free access to the super-computer built into the platform. |
| 20 | +If you become a regular user, you can connect your (super-)computer to BrainLife, which may significantly speed up your analyses. |
| 21 | +Before using Cairvoy, you will need a free Brainlife login. We recommend reading the [brainlife.io getting started guide](https://brainlife.io/docs/user/started/). |
| 22 | +There are also many helpful videos on the [brainlife.io YouTube channel](https://www.youtube.com/@brainlifeio/videos). |
| 23 | + |
| 24 | +Clairvoy uses the web-based ezBIDS tool (built into Brainlife) to import Dicom files, convert them automatically to NIfTI format, and then create a BIDS project (Brain Imaging Data Structure). |
| 25 | +You can learn more about ezBIDs at [this link](https://brainlife.io/docs/using_ezBIDS/) and watch a video at [this link](https://www.youtube.com/watch?v=KvhIHxzHsl4). |
| 26 | +The dMRI and T1 protocols can be found in Berto, et al. (Neuroimage: Clinical, In Press) 2024 and Nencka, et al. 2018 (Brain Imaging Behav 12:1121-1140). |
| 27 | + |
| 28 | +### Step-by-step directions for Clairvoy: |
| 29 | + |
| 30 | +A narrated video of how to use Clairvoy is available at [this link](https://youtu.be/TvVmsXITP_0). |
| 31 | + |
| 32 | +### Step 1. Upload files into a BIDS project |
| 33 | +1. Login into ezBIDS with your free Brainlife account. The homepage will now open to the `Upload Imaging Data` on the left side Tab. |
| 34 | +2. Drag and drop, or browse, for the files you wish to upload. Zip’ing the files may increase upload speed. |
| 35 | +3. Fill in the relevant details of the Dataset Description: Dataset Name (e.g., “DWI_workflow”, Authors (eg. your name). Some elements will be extracted from the DICOM, eg. Subject and Session. |
| 36 | +4. If the data is not deidentified, you may wish to change the subject name to a private subject ID number. |
| 37 | +5. In `BIDS Datatype`, choose `IntendedFor` to `dwi/dwi`. It is also best to deface at this stage, allowing the data to be shared in the future. |
| 38 | +6. Choose `QuickShear` method and run defacing (this will remove any recognizable face and ear features). When finished, visually confirm the brain is still in the image and select `Use Defaced`. |
| 39 | +7. In `Participants Info`, you may want to choose the sex and age of the subject (this is not HIPAA-protected information), then select `Finalize` to convert to BIDS.[^1] |
| 40 | +8. Next, click `Send to Brainlife.io` and select `Run DWI Pipeline`. |
| 41 | +9. Finally, click `Submit`, this will automatically create a Project on [brainlife.io](https://brainlife.io), and the pipeline will start processing the data inside the project. |
| 42 | + |
| 43 | +### Step 2. Switch to Brainlife |
| 44 | +1. Navigate to [brainlife.io/projects](https://brainlife.io/projects). |
| 45 | +2. In `Projects`, search for and choose your Project using the name selected in Step 1 (e.g., DWI_workflow). |
| 46 | +3. The `Pipeline` tab should show the pipeline is “active” and “processing.” Wait for process completion[^2]. |
| 47 | + |
| 48 | +### Step 3. Run the Clairvoy App |
| 49 | +After pipeline completion: |
| 50 | +1. Choose the `Archive` tab in your project, |
| 51 | +2. Select `tractmeasures`, and |
| 52 | +3. Choose `Stage to process`. Give the process the name “Clairvoy” and click `Submit`. |
| 53 | +4. In the `Preprocess` tab, choose `Clairvoy` and click `Submit App`. |
| 54 | +5. Choose the `Clairvoy App` from the list of Apps suggested by [brainlife.io](https://brainlife.io) and click submit. This should only take a minute or so to complete. |
| 55 | +6. When finished, the `Clairvoy App` output will state either: **This injury is at high risk of developing PPCS** in red letters, or **This injury is predicted to recover in under 28 days** in green letters. |
| 56 | + |
| 57 | +**[Here is a narrated YouTube video about how to run Clairvoy](https://youtu.be/TvVmsXITP_0)** |
| 58 | + |
| 59 | +[^1]: It is best practice to save the download `configuration/template` to your computer, which can be reused later for other similar datasets. |
| 60 | + |
| 61 | +[^2]: Time to completion depends on the currently available computing resources on [brainlife.io](https://brainlife.io). |
| 62 | +If you have access to a (super-)computer, time to completion may be reduced to less than an hour. |
| 63 | +If you use a free, public [brainlife.io](https://brainlife.io) supercomputer, the pipeline may take over 24 hours to complete. |
| 64 | + |
0 commit comments