How to Use 3D Slicer With Supervisely for Medical Annotation

To use 3D Slicer with Supervisely, install the ConnectToSupervisely extension from the Slicer Extension Catalog, log in, and open a labeling job from the platform. You annotate or review volumes in 3D Slicer, then submit them back, so your team keeps jobs, roles and review in Supervisely.

What is the 3D Slicer integration for Supervisely?

Supervisely has a medical labeling toolbox for DICOM, NRRD and NIfTI. Teams organize work through Labeling Queues or Labeling Jobs, which cover data preparation, pre-labeling, annotation, change tracking, activity logs and workforce statistics. This matters for CT and MRI, because the time of doctors is limited and expensive. The DICOM labeling docs describe the tool.

Some data workflows need other tools. For those, Supervisely integrates with 3D Slicer, open-source medical software for analysis and visualization of 2D and 3D medical images.

3D Slicer and Supervisely interfaces side by side

For more, see our guide to the Medical Annotation Toolbox and the 5-minute quickstart video tutorial on YouTube.

Is there a video tutorial for the integration?

Yes. This 8-minute video covers the most common pipeline for medical data:

• Workspace and data preparation with team organization.
• Creating a Labeling Job.
• The ConnectToSupervisely extension and its modules.
• Switching between the application and the web interface: start annotating in Supervisely, switch to 3D Slicer for 3D editing, then return to the web interface to complete the task.

Why use 3D Slicer with Supervisely?

3D Slicer is supported by an international community of users and developers with deep medical expertise. It was not designed to cover end-to-end computer vision workflows, so some functionality is missing. Supervisely adds:

1. Work in a web browser. If a device has no 3D Slicer, log in to Supervisely and edit data there.
2. Organization of annotation for a large team of ML specialists and domain experts.
3. Collaboration: user permissions, access control, external authorization (SSO, OpenID, LDAP), labeling and review tasks.
4. Several users annotating the same data at once, for consensus labeling.
5. Activity and performance tracking for annotators.
6. More export formats for training data.
7. Other annotation tools often needed in computer vision projects, such as tags and attributes.
8. Cloud storage support, so data is not only stored on annotators' computers.
9. Machine learning tools and neural network toolboxes for model training and pre-labeling.

	3D Slicer alone	3D Slicer with Supervisely
Where data lives	Local files on each computer	Platform storage, in the browser or in Slicer
Team workflow	Not built for end-to-end computer vision workflows	Labeling jobs with annotators and reviewers
Access control	Local application	Permissions and SSO, OpenID and LDAP
Annotation tools	Segment Editor and Slicer modules	Slicer tools plus the web DICOM toolbox

When to use 3D Slicer vs the Supervisely DICOM toolbox?

Use both in one job. 3D Slicer has evolved for many years and has many niche annotation solutions and visualization tools. Supervisely supports DICOM and other popular medical formats, and the annotation objects are 3D mask volumes, so 3D Slicer can work with this data directly.

Situation	Better fit
No device with 3D Slicer, or you only need to view or edit data	Supervisely in the browser
Quick first mask, for example a lung	Smart Tool in the Supervisely labeling tool
3D editing, niche modules or Slicer's visualization tools	3D Slicer with the extension
Many annotators, roles, review and activity tracking	Labeling Jobs, from either tool

An example of the most common pipeline for working with medical data.

Supervisely adds AI-powered and automation tools that speed up labeling of 3D volumes.

How to use 3D Slicer with Supervisely step by step?

Supervisely released the ConnectToSupervisely extension for the 3D Slicer Extension Catalog. To use it:

1. Import data and prepare the workspace: create a project, import data, invite annotators.
2. Create a Labeling Job and assign it to annotators and a reviewer.
3. Install the extension in 3D Slicer.
4. Open the extension module and log in to Supervisely with the account that has the task.
5. Select the Labeling Job, click Start Labeling, annotate and submit the result to Supervisely.
6. Review the annotations in the reviewing module.

ConnectToSupervisely extension module workflow

The extension has two modules:

• Labeling Jobs Annotating for annotators to do the actual labeling
• Labeling Jobs Reviewing for reviewers to validate and correct annotations.

Labeling Jobs Annotating

In terms of user interface, these modules look the same with little difference in functionality.

Step 1: How to import data and prepare the workspace?

For more on the platform, see the video course.

You need an account. For the free version, sign up for the Community Edition. Enterprise Edition users have their own platform address.

After you log in, you become a member of an automatically created team, likely named Team of <your login>, with the default First Workspace selected.

• Within a Team, users share resources such as projects and models. Projects can only be created within a team, and you can switch teams or create new ones from the left menu.
• A Workspace is a separate "room" in a team where users organize projects and datasets. Each team has at least one.

You land on the Projects page.

• A Project is a combination of datasets and meta information (like classes and tags). It is the main building block of data organization in Supervisely.
• A Dataset is like a folder where your data is stored.

Click Import Data and follow the instructions. Then create the Classes and Tags you need for annotation.

Next, add annotators to your team. In the lower-left corner, click the name in the Current Team section and go to Manage Members. In Team Members, click INVITE, type the login and select a role, for example Annotator.

Inviting a team member in Supervisely

Step 2: How to create a Labeling Job?

The extension is based on Labeling Jobs, so you need one before you start.

Labeling Job in a Nutshell

Labeling Jobs organize and distribute annotation tasks within a team. Annotators work on well-defined portions of the dataset and follow consistent guidelines, which keeps data quality high.

Read more in "Mastering Labeling Jobs: Your Ultimate Guide".

Go to Labeling Jobs in the left sidebar and click + NEW in the upper right corner.

At minimum, describe the task and select the annotator, a reviewer, the project, and the classes and tags available for annotation.

Step 3: How to install the 3D Slicer extension?

The extension works on Linux, macOS and Windows. We recommend installing it from inside 3D Slicer. There are two ways:

• From the Extension Catalog
• From a local archive file, downloaded from the extensions catalog or from the GitHub releases

Install from the Extension Catalog in the application

☝️ If you get an error when you click on the "Install" button,

sometimes updating the catalog using the "Check for updates" button helps.

Install from the downloaded archive file

Files for each OS are on the extension catalog website and in the GitHub releases. Select your operating system in the top right corner of the catalog page.

• In the Extension Manager Wizard of 3D Slicer, select Install from file.
• Select the downloaded file and click Open. After installing, you are prompted to reboot the application.
• After rebooting, the first time you open a module, an information window shows the installation of the Supervisely Python SDK, which the extension needs. Wait until it finishes.
• 3D Slicer then reboots automatically, and you can start working.

☝️Each release is marked with the 3D Slicer version it is built for. Installing on a different version may fail. In that case, use the Extension Catalog option.

Step 4: How to open the extension and log in?

After the install and restart, open a module by selecting the Supervisely tab in the list of modules.

Pro Tip
To get quick access to the modules:

→ Open the Edit menu on the top bar

→ Go to the Application Settings
→ Choose the Modules section
→ Find in the Modules unit Supervisely's modules
→ Drag and drop them to the Favorite Modules unit and save with the OK button

Authentication Tab

Log in to the Supervisely server where your data is located.

Enter credentials here

With the Remember login setting, the extension does not store what you typed. Later logins use the received API token, which is stored in a separate .env file in the current user directory of the operating system.

You will be automatically logged in

Teams and Jobs Tab

After login you see the Labeling Jobs available to you in your teams.

Team and Jobs

The list shows only Jobs with status Pending or In progress where you are the annotator. Click Refresh Jobs List at any time to update it. This does not reset your current workflow.

Active Job Tab

In the Labeling Jobs Annotating module, the Info tab shows the Job description and the list of Volumes to annotate. You can edit Tags, set statuses for Volumes, see how many Volumes are annotated, and send the Labeling Job for review when the work is done.

Labeling Jobs Annotating: Active Job

In the Labeling Jobs Reviewing module, the tab is almost the same. It lists the objects to check with their annotations. You set statuses for Volumes, then restart the Job or finish it with the result.

Labeling Jobs Reviewing: Active Job

☝️ In both modules you create and edit annotations with the built-in 3D Slicer tools: Segment Editor, Segmentations and others.

Settings Tab

This tab configures automatic triggers and other data-saving options. New options will appear here.

Step 5: How to annotate volumes in 3D Slicer?

In Labeling Jobs Annotating, select the Team and Job. Click Start Labeling to download the data from the server.

💡 Speed up your workflow by starting in the Supervisely Labeling Tool and then switching to 3D Slicer. For example, to mark a human lung quickly, use the Smart Tool:

Smart Tool Annotation Process

Learn how it works in "How To Use Model Assisted Labeling To Boost Annotation Performance With Supervisely".

If you used the Smart Tool, click Sync Job to get the latest changes from the project.

When the data is downloaded to the Working directory (adjustable in Settings), the Active Job tab unlocks. Selecting a Volume loads its data automatically.

In the module, you get:

• A list of available and attached Tags for the Volume

On the scene, the extension creates:

• Segmentations, which are the Classes available in the project
• Segments, which are objects of these Classes

Segmentations and segments of a volume in 3D Slicer

• 1 Volume: CTChest.nrrd
  
• 2 Supervisely Class: Tumor → Slicer Segmentation: Tumor
  
• 3 Supervisely Annotation Object: Tumor → Slicer Segment: Tumor
  🤓 object name is always the same as the class name

Annotating

To edit, create or delete an annotation object, use any other 3D Slicer module that works with objects on the scene, such as Segment Editor or Segmentations.

To create a new object, select a Segmentation and add a new Segment. You do not need to name it or change its color, because it is converted to the class name and color on upload. You can still rename or recolor it for convenience.

Creating a new segment in 3D Slicer

In Segment Editor you can use the Ignore Segment status on Save setting to control saving more flexibly.

Tagging

• The Tags tab is not active if there are no tags.
• Tag availability depends on the Labeling Job settings.
• Tags can be added ➕ and removed ➖.

Click a tag in the list of available tags to attach it to the Volume. Click an attached tag to detach it.

Status changing

After you create the annotations and save, set the "Done" status for the current Volume. You can still make changes afterward.

When all tasks in the Labeling Job are complete, click Submit for review. The Job status changes, it disappears from the list, and the scene and local Job data are cleared.

Step 6: How to review annotations in 3D Slicer?

Review works like annotation, with extra statuses, the ability to Restart or Finish the Labeling Job, and its own module settings.

The reviewer usually:

1. Reads the technical description and opens the Labeling Jobs Reviewing module.
2. Optionally corrects the labeler's annotations.
3. Accepts 👍 or Rejects 👎 each Volume.
4. Finalizes the Labeling Job once all data in the dataset is reviewed.

💡Extension settings explanation

Example: Annotating Module Settings

Remember login

The server address and the account token from the first login are saved to a .env file for simpler connection later.

Ignore Segment status on Save

When checked (default), all segments are saved regardless of status. Otherwise, segments with status Completed are saved, and segments with status In progress initiate a save request.

Segments with other statuses are not saved, but their state in the current scene does not change. To reset it, use Sync Job.

⚠️ Deleted segments are deleted on Save regardless of settings and statuses.

Segment statuses

Auto save when changing Volume

Saves your changes when you switch between Volumes without saving manually.

Auto save on "Submit for review" Annotating module only

Saves automatically when you submit the Labeling Job after the last Volume. To submit without your latest changes, uncheck it first.

Restart Job with only rejected volumes Reviewing module only

Two options for restarting the Job:

1. By default, it restarts the Job only with Volumes marked Rejected.
2. When off, it marks all unmarked Volumes (neither Accepted nor Rejected) as Rejected and restarts the job like the first option.

Working directory

The local folder where Labeling Job data is stored.

Restore Libraries

If Python library versions conflict during the Supervisely package installation and you still want to proceed, you can restore the libraries to their original versions.

Summary

The ConnectToSupervisely extension lets you work with platform data from 3D Slicer as easily as from the web interface, and still use the extensions in the 3D Slicer catalog. It is open source, so you can also extend it.