Master Troubleshooting Guide

​

Overview

This document is regularly updated with new entries, both for public reference and to further improve the chatbot. If you encounter a problem on the platform that is not covered here or a bot-produced output that resembles a hallucination, please reach out to us at support@redbrickai.com.

Big thanks from the RBAI Support Team!

​

Login-related issues

Q: I’m encountering an “error while signing in” error, but I know my password and username are correct.

A: If you’re 100% sure that your credentials are correct but cannot proceed beyond the RedBrick AI login screen after clicking on “Sign in”, you may have to clear your cached site data associated with the RedBrick web app.

You can do this by adhering to the following steps:

Q: I tried resetting my password but I’m not receiving a confirmation email.

A: If you have initiated a password reset but have not received a confirmation email, there are several recommended steps you can take:

Q: I have Email MFA activated but I’m not receiving a code via email.

A: A: If you are not receiving your MFA confirmation code via email, there are several recommended steps you can take:

​

Compliance, Information Security Assessments, etc.

Q: Who should I contact if I want to discuss any of these:

• Security policy, architecture, or reporting
• Data privacy policy, architecture, or reporting
• Compliance posture

A: Please reach out to us at support@redbrickai.com.

Q: What compliance frameworks and regulations does RedBrick AI and its affiliates adhere to?

A: SOC II Type 2, GDPR, and HIPAA. Clients who subscribe to annual plans have access to SOC II reports, as well as various other types of security reports.

Please reach out to support@redbrickai.com with further questions and concerns.

​

Data Privacy and Architecture

Q: Does RedBrick AI download our image data?

A: It depends, but generally no. If you utilize AltaDB or Direct Upload when importing Tasks to RedBrick AI, then those images and all accompanying files (such as text reports) and metadata are stored on Redbrick AI’s servers.

If you have integrated an external storage method, such as AWS, Google Cloud, or Azure, then your image data, metadata, and accompanying files (such as text reports) are not stored on RedBrick’s servers.

In that case, the client-side (i.e. the computer that your colleagues use) issues all of the requests to your file storage, circumventing RedBrick servers entirely.

In other words, when one of your users clicks to open an image on RedBrick AI, they request the file temporarily from your bucket to their computer.

Please note that all segmentation files are stored on RedBrick’s servers by default in NIfTI format. You can optionally choose to store all segmentations generated in RedBrick in your external storage as well by designating the storage method on the Annotation Storage page in Project Settings.

​

SDK Issues

Q: I ran put_tasks(existing_labels=True) while attempting to upload labels to RedBrick, but I can’t see anything in the Editor! My Tasks are empty, i.e. I can’t see the labels I wanted to upload.

A: The put_tasks() function is used for two operations:

1. Moving a Task(s) further down a Project pipeline (i.e. submitting Label Stage Tasks, or accepting or rejecting Review Stage Tasks)
2. Uploading labels to an existing Task

In your case, the existing_labels=True parameter will simply submit the Tasks in their current state on the RedBrick platform, preventing you from uploading any labels you may supply in the upload JSON.

If you wish to upload labels that you are supplying in your current code to existing RedBrick Tasks, you should be using existinglabels=False

​

Workflow Issues

Q: I need to remove a Project. What are my options?

A: Assuming you have proper permissions, you can either:

• archive a Project or Workspace
• delete a Project or Workspace

Archiving a Project or Workspace can be done within a Project or Workspace’s settings. This will move the Project or Workspace to a special “Archived” Section of the Dashboard and prevent users from accessing it.

If any member of your Organization wishes to access an archived Project or Workspace, they’ll have to restore it by first clicking on the Project or Workspace and then clicking on the corresponding Unarchive button.

A Project or Workspace can be deleted by clicking the corresponding Delete button in the Project or Workspace’s Settings.

Projects can also be deleted programmatically using the SDK.

​

Data IO Issues

Q: Why is there a “binaryMask": True value in my export tasks.json file? I didn’t use the binary_mask parameter when doing the export.

Related Q: Why does my segmentation export have a long list of instance-1.nii.gz, instance-2.nii.gz, etc., instead of the usual single NIfTI file?

A: RedBrick AI will automatically export segmentations that overlap as a binary mask, generating one instance per class.

Compare the two JSON blocks below:

Code Block 1: a standard (i.e. non-binary mask) export, containing a single NIfTI file.

Each NIfTI file contains all of the individual segmentations associated with a particular Series as defined by the segmentMap, which can be found in the tasks.json file.

projectId  
|-- segmentations  
      |-- Task_001 // Folder with the name of the Task (i.e. "Datapoint" on the Data Page)  
            |-- SeriesName.nii.gz  

Code block 2: a binary mask export, containing:

• an aggregated annotation file containing all of the segmentations in the Series;
• a subdirectory that has the same name as the Series;
• within the subdirectory, an individual annotation file for each segmentation;

Please note that the annotation files in the subdirectory are generated incrementally based on the order in which annotations were created by the labeler.

projectId  
|-- segmentations  
      |-- TaskName  
            |-- SeriesName.nii.gz // NIfTI file containing all segmentations  
            |-- SeriesName   
              |-- instance-1.nii.gz // NIfTI file for first overlapping segmentation  
              |-- instance-2.nii.gz // NIfTI file for second overlapping segmentation  

Q: I want to show a custom value (such as a custom tag) in the viewport so the annotator can see it. What are my options?

A: You have two options: adding Task or Series-level metadata or uploading a separate document to our document viewer.

Use Task or Series metadata when:

• you want to have Series-specific information available in the same viewport as the image;
• you are working with one DICOM Study per RedBrick Task (i.e. any Task metadata that you upload will correspond to every scan inside of the Task);
• the annotator doesn’t need to have a constant visual reference (e.g. a 2x1 view with the image in the lefthand viewport and a text report in the righthand viewport);
• the information you’d like to provide is easily categorized and relatively short (e.g. DICOM spacing tags, header-based information, a one-line diagnosis);

Use a Report when:

• your annotator needs to have constant access to certain information about a specific scan without toggling it on or off;
• the information you’d like to provide is more comprehensive in scope (e.g. a full clinical findings report, patient history, etc.);

​

How to Format Task or Series Metadata

[
  {
    "name": "Task Name",
    "series": [
        {
          "name": "Series 1",
          "items": [
              "path/to/file.dcm"
          ],
          "metaData": { "Spacing (0018, 0050)" : "1.0" } // this is Series-level metaData with a single entry
        }
      ]
    "metaData": { // this is Task-level metaData with multiple entries
      "Study Description (0008, 1030)" : "CT Abdomen and Pelvis w/Contrast",
      "Study Date (0008, 0020)" : "20250415" 
    }
  }
]

​

How to Format a Text Report

[
  {
    "name": "Task Name",
    "series": [
        {
          "name": "Series 1",
          "items": [
              "path/to/file.dcm"
          ]
        },
        {
          "name": "Text Report",
          "items": [
            "path/to/text/report.txt"
          ]
        }
    ]
  }
]

You cannot add custom tags or other forms of metadata using Hanging Protocols.

Q: I deleted a Task from the RedBrick GUI and tried to re-upload the file, but I’m getting an error in the Terminal that says a Task with this name already exists. What can I do?

A: Attempt the upload again using the --clear-cache flag.

Q: I attempted an export and got the following error: Failed to process segmentations: Invalid label files - what can I do?

A: This error most commonly happens when RedBrick does not have the proper permissions to fetch (GET) or add (POST) annotation files from or to your external Storage Method.

You can refer to the following guides for more detailed instructions on how to verify your bucket’s configuration:

• Amazon s3
• Google Cloud
• Azure Blob

As always, you can also reach out to support@redbrickai.com for further assistance.

Q: I want to export a previous version of a Task’s annotations.

A: If you need to export an older label set (i.e. a label set that is older than the Task’s currently designated label set), you’ll need to use both the UI and the SDK/CLI.

​

Billing

Q: If I have a question about my billing, who do I contact?

A: Please reach out to us at billing@redbrickai.com.

​

Task loading

Q: My task is taking a long time to load, is there anything I can do.

A: Typically the best solution to all of these problems is to make use of the Worklist feature and the “Cache Task” action. These will download the images locally so that when the user opens the task they will be able to work right away.

There are many reasons why data can take a long time to load. The following are some possible problems:

• Slow network speed - use a internet connection speed test. If your speed is under 10 MB/s then you will probably experience some issues using RedBrick. For teams working with large tasks, speeds upwards of 50-100 MB/s are recommended.
• DICOM compression format - DICOM images have different compression formats, some of these take significant amounts of effort to decompress and this can slow down rendering your images. You can try re-encoding your images with a different format. If network speed is not a limitation, uncompressed images will load the fastest.

Q: My task is stuck loading and never actually shows me any images or error messages

A: While we try to prevent any undefined or hanging states such as this, it may occasionally occur. We recommend the follow steps:

1. waiting 10 minutes just to be sure it is not caused by a slow internet connection
2. Ask a colleague to open the task on their computer, it is possible that some computers may not be able to open all tasks due to resource limitations.
3. Download the task files locally and open them in the Preview Tool. Check for errors with the files or loading.
4. Reach out to support@redbrickai.com. Let us know what you’ve tried, and share as much info as you can about the images. Including URLs to tasks that won’t load is helpful for us to quickly pin point and solve any issues.

​

Working with Segmentations

Q: I see a red area on my segmentations that I cannot edit or delete, what do I do?

A: This can occasionally happen from user uploaded segmentations or other anomalous events. It is caused by untracked instance ids in the underlying pixel mask. This can be fixed by running the “Detect Untracked Segmentations” action in the command bar. This action should convert the red untracked instance ids into unclassified segmentation objects. These can then either be deleted, have their class reassigned, or merged into another entity with the boolean operation tool.