Tutorial: New Project

This tutorial will guide you through creating a Datamint project, adding images, setting up an annotation workflow, and adding collaborators. It is intended for principal investigators.

A Datamint project is for developing a model or set of models that perform a single medical task. For more information, see What is a Project?.

For this tutorial, we will create a hypothetical project which interprets ultrasounds of elbows. We will not actually be training a model, but we will do everything neccecary to begin training one.

Creating the Project

To start, go to app.datamint.io and sign in. Click the Add Project button.

_images/AddProject.png

Give the project a name and description, and click Create Project. We aren’t going to be adding files here, since we don’t have any uploaded yet.

_images/NewProjectScreen.png

Open the project. You should see a screen like this:

_images/ProjectSummary.png

Adding Images

We will now upload some images for annotation. Start by going to the Files tab.

_images/AddFiles.png

We now need to upload some data. Let’s use a collection of open-source elbow MRIs. You can download it from the Datamint website here. [link]

Click the Upload Files button. You should see this menu pop up.

_images/UploadFiles.png

We can also set tags that will be applied to the images. Let’s give all these images the “elbow” tag: to do this, click the “Add tag” box, type “elbow”, and press enter. You can make a tag with any text you want.

Creating an Annotation Workflow

Now, let’s set these images up for annotation. [Setup stuff]

We can now see different sorts of annotations which can be applied to the image. For this tutorial, we’ll be using all of them, but your AI model most likely only requires some of these to be used.

Creating Segmentations

We will create a segmentation for the posterior fat pad.

_images/AddSegmentation.png

Click “Add Segmentation”, and a new segment called Segmentation 1 will be created. Double-click on its name to edit it.

_images/RenameSegmentation.png

Creating Labels and Tags

Labels and Tags allow for data to be classified. They can apply to a whole image, or to individual frames.

Labels are

Creating Shapes

Shapes (such as lines, rectanges, or circles) indicate specific coordinates to the model. Each shape you create can only be used once per frame.

We will mark the basilic and cephalic veins with lines. Since there are two veins to mark, we need to create two lines. Press the Add button in the Shapes and Regions row, select Line, type “basilic vein”, and press OK. Repeat for the cephalic vein.

_images/CVeinLabel.png

Adding Collaborators

Stub

Making Annotations

Stub

Reviewing Annotations

Stub