Annotation

Annotation Project enables the AI developer and radiologist to curate datasets to develop or test ML Models.

CARPL comes in-built with a fully-functional DICOM Viewer with an annotation platform that allows radiologists who are either reviewing the findings of an AI algorithm, or independently testing an AI algorithm, to document their findings in a simple yet comprehensive way. The radiologist can easily edit, add or delete the pixel level annotations within the CARPL viewer and save changes

CARPL also gives AI developers and radiologists the ability to define input text fields for radiologists to fill as they go about the validation process. This is extremely critical for enhancing communication between the radiology and data science teams since often there are comments that the radiologists might want to share with the data science team.

The annotations performed by radiologists can be downloaded as JSON files by the AI developers and incorporated into their development pipelines.

Create Annotation Template

Click on "Annotation - Annotation Template" from the left panel
Select "Create Template" from the top right corner
Provide "Template Name, Template Description, and ROI Labels "
ROI Labels can be selected from the dropdown menu or can be created .
Write content in the ROI labels form control and enter space to create a new ROI label
Drag elements from "Form elements" boxes and drop it in the form builder box
Edit the "Form elements" as per the requirement.
User can select "Required" to mark mandatory.
Select "Preview" to review the template before saving the template
Select "Save Template" to save the annotation template

Show Annotation templates summary

Select "Annotation" -> "Annotation Templates" to view all the templates created by the logged in user
Select "View Template" to view individual templates created by user
Once template will open, click on "edit" icon to edit form controls
Use "Delete" icon to delete any form control
"Save template" to save the updated template.

Create Annotation Projects

Click on "Annotation - Annotation Projects" from the left panel
Click on "Create Annotation Project" in the annotation platform
Provide "Project Name", "Description" and "Select Annotation Template "
Select "Proceed" then select desired "Dataset" from the datasets drop down
Select "Proceed" to create an Annotation Project
Annotation project will be created and user can see the newly created project in the Annotation summary Panel

Show Annotation Project Summary

Click “Annotation -> Annotation Project” to create, view, edit and delete annotation project..
Use “View Template” icon to view the template
Use "Download annotation ROI” icon to download ROI level annotation in a json file
Use “Download annotation ” icon to download study level annotation in a csv file
Use "Download Logs" icon to download annotation logs in csv file
Use “Assign” to assign cases of a project with other users
Use “Edit project” icon to edit the project detail like project name and description
Use “Delete project” to remove a project
Click on "Annotate" button to annotate each scan of the project
Click on “Review” will allow the assigner to view all the cases present in the project. Only the assigner has access to review all cases annotated by all the annotators. The review button is only available for assigner of the project.

Assign Annotation Projects

Once annotation project got created click on "Assign Annotation" icon to assign annotation among users.
Select studies and then click "Assign Annotator ", to assign annotator(s). Write email ids of the annotators click "assign".
Select "Assign to Myself", to assign studies to the logged-in user for annotation
Select studies and then click " Remove Annotator", to remove users for annotation.
The annotators will be notified via email.

Review Annotation Project

Assigner can review status of the annotation assigned to other annotators
Select annotator's name from the annotators drop-down list to review the status of individual study
Select "view Study" to see the ground truth by Radiologist
Use the "Completed" or "Pending" filters to visualize studies based on status filter.

Annotate with CARPL Viewer

Once assigned, Click on "Annotate" to start the annotation.
Clicking on "Annotate Study" will open CARPL annotation platform "CARPL Viewer"
CARPL Viewer will open the selected study
Annotators can mark with the help of desired ROI tool from the left panel and mark the findings
AI findings can viewed and updated through "AI Panel"
"Save and Mark Complete" will mark annotation status as "Complete".
Mark "Auto-save" ON to automatically save the annotation
Select Next or previous button to annotate immediate studies

Page updated

Report abuse