GSoC Week 4

This week was hectic. I continued learning a java and understanding the components of the LibreHealth Radiology module. I spent my time exploring tutorials, studying the code of the LH-RIS (LibreHealth Radiology Information System), and immersing myself in Orthanc documentation. For context, Orthanc is the PACS (Picture Archiving and Communication System) or in simpler terms it stores the X-ray images and has methods to communicate and share them.

Architecture for better understand,





Challenges and getting past them,

Last week I coded the new properties to enable users to include model information in the admin settings for model selection. I added code in all the pages I found needs adding but when I pushed the changes to my forked repo, there was a pipeline failure. I could not figure out why. I looked at other older commits who did the same and I could see in recent commits developers adding configurations to only the config.xml. I did the same and it worked.

I created a method that pulls information on all the available models in the AI model service. The output of the method will be used in the administration settings page to fill in the model task properties dropdown. Right now it is a text box field and will be converted to a drop down. I was able to implement the method but I got stuck on how to make the admin page model properties textbox a drop down and connect it to the newly created method to pre fill it. I have not figured this out yet and I will clarify this with my mentor.

Moving on to the next challenge. So the task is to do model selection by using the DICOM image metadata. We need to read the information in the metadata and call the appropriate model. For example if the DICOM metadata has information on chest X ray then we call the chest x ray AI model. For this I will need to read the DICOM metadata from the Orthanc server, read the model properties for all models from the administration settings and send them to the the AI Model Service where the service will look at the metadata information, filter in the appropriate model properties and redirect the corresponding AI model endpoint.

I faced two issues here,

One, how do I read the DICOM metadata? After rummaging through the code for a while I found Utils.java page that creates a URL to pull the DICOM image from the Orthanc Server. Now I need to do also pull the metadata of the DICOM image. I started going through the Orthanc documentation to figure this out but there were a lot of terminologies I did not understand. After a while I figured out  they use an attribute "metadata" to pull the metadata information. I spent a lot of time to just figure this out. Then I went on to implement the methods.

Two, there is no common pattern in the DICOM metadata to separate out a chest image from a Mammogram image. For example, for a chest image you may need to look at the DICOM metadata attribute "BodyPartExamined" which has value "CHEST" but for Mammograph the attribute would be "Modality" with value "MG". The problem arises now, how can you without hardcoding different attributes for different modalities (chest, head, mammogram, etc) do model selection? I need to talk about this to my mentor and get it clarified.

Last I modified the existing call from the LH-RIS to the AI Model Service to include DICOM metadata information and model properties from the admin settings page. After making the changes I wanted to test if the data sent from the method to the AI model service was correct and working. I need to understand here how to create test scripts but for now I made a change in the AI model service to which the LH-RIS module pushes the data to, saved it and checked the output. This way I was able to test and correct my code. I need to talk to my mentor on efficient ways to test these things.

Finally,


Through all these challenges I was able to make progress and pushed merge requests. I could not meet with my mentor this week since he is traveling a lot recently. I know I will meet with him soon though, align and progress on my work.

Comments

Post a Comment

Popular posts from this blog

Google Summer of Code 2023: AI Model Service Auto Selection by Modality - Final Report

Image
Introduction GSoC is one amazing program that lets students and professionals get into the world of open source, contribute code and also learn back from the program. I definitely learnt a lot by contributing to my mentoring organization LibreHealth through GSoC'23. I now understand the value of open source and will become a lifelong contributor wherever and whenever I can. LibreHealth is an umbrella organization where multiple health information technology projects collaborate. I was actively involved in contributing code to their LibreHealth Radiology application  through the project "AI Model Service Auto Selection by Modality". My contributions to the application were more involved into implementing AI models in the application and also building modules that do automatic model selection, example, when a Radiologist opens up a Mammography image the application automatically triggers the Mammography AI model to produce inferences on the image. Objectives The project des
Read more

GSoC Week 1

This week I was researching AI models to implement for the different modalities. More specifically, Mammography. I also finally had the opportunity to talk to my mentor on Thursday (June 1st, 2023). You can find the recorded meeting link  here . Building a Breast Cancer Classification Model from Scratch At the start of the week, I attempted to implement a VGG16 classifier to distinguish between malignant and benign calcifications in a mammogram. Using PyTorch, I developed the model and achieved a validation accuracy of approximately 70%. The source code can be accessed  here . This model and code serve as a starting point for implementing a more complex model mentioned in this  paper . Briefly, the paper discusses building a model solely on the region of interest (ROI) of the calcification/tumor and adding layers and training on the entire images. This method has significantly reduced the training loss and, consequently, improved overall accuracy. Mentor Discussion and Change in Direct
Read more

My First GSoC - LibreHealth Community Bonding Period

GSoC Acceptance I was amazed to know that I got accepted into Google Summer of Code 2023. This is my first time applying and getting into GSoC. I remember the time when I sat at the library for hours, working on the preliminary tasks, figuring out solutions, and writing an extensive proposal. Reiterating through the solution and proposal back and forth, I watched lots of YouTube videos and read blogs providing tips to draft a good proposal. All the efforts paid off when I received the email of acceptance. The Org & Project I drafted a proposal to the LibreHealth Organization. LibreHealth is collaborative community for free & open source software projects in Health IT, and is a member project of Software Freedom Conservancy. It is amazing to just think about how contributors from all over the world come together and build things that impact the world positively. This idea is what brought me into open source. Before jumping into the project details, here are a few things about m
Read more