Project: Image Annotator for Radiology Images

Community Google Summer of Code
, ,
1

Summary

Develop a web-based tool to annotate ROI for radiology images. It could be either bounding box annotation or pixel segmentation.

Deliverables:

As a student, this year task should be :

[1] To edit the region of interest in the DICOM viewer.
[2] To import the studies from other DICOM viewer, so that we ensure compatibility standards between DICOM Viewers and make it available for the radiologists to diagnose.
[3] To export the details of the key object or Region of interest we created on the DICOM study/series, so that we have a common file that can be passed on along with the DICOM images to other DICOM viewers.

Prerequisites : web development, DICOM

More details will follow soon. :slight_smile:

Mentors : @geepriya @judywawira

Find attached files for more reference.

usecase.pdf (196.8 KB) Presentation.pdf (1.8 MB) findings.pdf (341.6 KB)

1 Like
2

Hello!! My name is Ankita Kulkarni. I am pursuing my Bachelor’s Degree in Computer Science and Engineering. I am in 4th year. I have experience in image processing, in it more specifically work on particular Region Of Interest and also image segmentation. I am looking forward to work under this organization and want to contribute in open source project for LibreHealth under gsoc2020. What are steps I should follow for further procedures?

1 Like
3

Hi @pri2si17, Thank you for outlining the image annotator project. Should we develop a POC and provide a link to the Github repo for this project as well?. Thanks.

4

In your proposal and ideally here. You should be getting feedback before you finalize your proposal.

5

Hi @r0bby, Thanks for clearing that up, absolutely will do.

6

Hi @pri2si17 @r0bby I am interested in working for this project as I have background in image processing . I wanted to know if we are supposed to contribute to any repo before starting the project also what are the steps before submission . Thanks !

7

Read the thread. Develop a POC, just a basic UI. The link to your github/gitlab repo should be here and in your proposal.

8

Hi @pri2si17 @r0bby ! I am Deepak Singh from India. I am interested to work in this project. I have some experience with image processing. Thanks all.

9

Hello Everyone, To kick off with this project, [1]. You may explore algorithms that bounds/segments the area/region of interest in the DICOM viewer and make it flexible to be edited by radiologist. Technically, using an HTML Overlay over the image.

You should be able to view a basic prototype using the below link https://xd.adobe.com/view/816d8435-a046-4c3b-6388-450763493e4d-f75c/

[2] & [3] You may explore and come up with ideas to implement the import, export feature of the annotated/edited image from the first requirement [1]

This is the repo on gitlab where you can see the presentation notes, images, reference links that can help you get started

Feel free to reach out if you have any queries.

10

Hi, I’m Abhinav Srivastava, few questions.

  1. Is this to be integrated in LH-Radiology?
  2. Do you expect a ground up work or can I use tools like Oviyam?
  3. Is this hosted online or is it a locally run service?
  4. “To import the studies from other DICOM viewer”, is this a metadata import or an HL7 communication?
  5. Between accuracy and speed, which is prioritised for RoI?
11

If you are just joining, I’m afraid it’s too late to apply.

12

Is that true for all projects? Or this one specifically?

13

All of them. The coding tasks take time to get correct. If you want to guarantee we don’t ignore it outright, submit PoC code…otherwise, i’m afraid it’s going to be ignored.

Doing it correctly at this point is going to be hard.

14

This project seems to have no coding challenge: Project: Incorporate reporting workflow into Radiology

I have already worked on similar projects, as long as I know I wouldn’t be discriminated on basis of this, I would continue with this.

15

There is, it’s just not defined clearly – which is an issue