Project: Incorporate reporting workflow into Radiology

@r0bby hello i hope you are okey, i want to ask you , i read the whole user stories and read the almost all codebase before completing my proposal i want to understand, right now how many parts of user stories completed mentioned at lh-radiology-user-stories gitbook. because it looks MRRT templates ready so we will be building RadiologyStudyServiceImpl, RadiologyReportService and RadiologyRestController? and other missing parts. And what will be Dictation software?

Kind Regards.

@judywawira or @ivange94 are your best bets – they will answer here – DO NOT pm them please.

ok so i will be waiting here, thank you very much and i know no private pm.

The workflow described in this project is not yet implemented but yes MRRT templates are already supported.

so we will be building RadiologyStudyServiceImpl, RadiologyReportService and RadiologyRestController? and other missing parts.

I’m not sure what you mean by that. Submit a draft proposal on the gsoc website and I’ll have a look. You don’t need to worry about the every user story except these ones Residents · lh-radiology-user-stories

2 Likes

thank you, i guess i have misunderstanding but thank you for lead. i will upload as soon as possible. and i want to ask in objectives 3rd objective is " * Support workflow for managing MRRT templates for each user or a group of users" what does managing MRRT templates for users mean here?

Kind Regards.

1 Like

LibreHealth GSoC Proposal.pdf (145.8 KB)

@ivange94 Sir can you look at my proposal, please provide my feedback because i had mistakes i want to fix them before Students Apply. Any help appreciated.

Kind Regards, Oguz

@r0bby I am proposing an idea. Why not we detect the areas on the medical image which can be a some deviation from a normal body radiology. That will help the radiologist to give attention to the area easily. And report can be made very fast. Or, based on the outcome of the detection of abnormality from X-ray, we can make a machine generated report and attach that with the original report. How is the idea sounding?

How would you do this @Mainak431?

@judywawira See above.

@r0bby I will use deep learning to detect the area of interest from the image. And among those area of interests we can detect the abnormality. If abnormality is above some threshold value. We will mark that area as red. And generate report as - Crack detected - probability - 79 % or something like that.

I will let @judywawira decide on the merits.

@Mainak431 you are welcome to write your own proposal and we will evaluate it … There are other deep learning projects that you could easily work on – go to the GSOC project listing

@judywawira @r0bby I will like to propose an idea along with the previous one for user side of the report. Like, in the medical report common man often find terms which they never heard of. So, every such term will have a quick help link. Which will link to be most searched page for that link using google api. I think that will be very useful for the user end. Also we can add some lines related to that term in a small paragraph just below that term. What do you think?

@Mainak431 – Your proposal will be held to a much higher standard, so it better be detailed.

Good luck.

@r0bby there is one issue here. Maximum machine learning implementations are in python. Tensorflow is not stable in Java . If i want to integrate this AI based features in JAVA, i will need a python like space or a communication between python and java. Communication means data transfer will be required. Although Py4J and Jython is available for data transfer. But I would like to know if you currently have any such interface?

I think there is a workaround though to use event driven programming to detect file changes and executing according to that event in both python and java side. @r0bby . Please let me know if you will allow python code in your project or not.

Discuss the drawbacks of all of this in your proposal. You need to discuss how you will implement this and the benefits and drawbacks of each.

@r0bby Please look at my proposal and provide feedback. https://docs.google.com/document/d/1s4Q3W31sugBzvJEopngeQ9mDZtLeeNbKjC7Ih8qfnVc/edit?usp=sharing

  1. You have some careless typos in here.
  2. We don’t need references.
  3. No wireframes whatsoever, how will this look?

@r0bby Looks depend on your frontend design. After uploading the image, an additional window will open side wise depending on your frontend design and the image will have red rectangles wherever some abnormality is detected. And similarly for report help or the second part, on just tapping the mouse on the Term, a similar window based on your front end design will open and search results will be shown.

That’s not how this works. You are proposing a system. You need to know show us some kind of a mockup as to how this will work. Remember, since you are proposing an idea – it needs to be robust and complete. That includes thinking of how the UI will look and operate.

have a look at https://radiology.librehealth.io – how would it work if you were to implement your project now? Would it require new systems?