GSoC 2026 Introduction and Questions regarding project(#13)

[aasthathakkar]

Hello, Diomede/Alaska team and respected mentors(@KathiraveluLab, @bananthreddy30)
 
I am Aastha Thakkar, a CSE student IIIT Vadodara, from Gandhinagar, Gujarat, India(IST timezone). I have been working with python and web development technologies(like Js, flask etc) through academic and personal projects. I also explored and learnt basics of Machine Learning and Generative AI, which drew my interest towards healthcare data systems, especially medical imaging workflows.
 
I came across Alaska while exploring GSoC 2026 organizations, and focus on improving accessibility and usability of medical imaging data made me compelled. The project "Creating shareable albums from locally stored DICOM images'(#13) caught my attention. Also, building this using the listed tools(like Kheops, MEDIator, or OHIF) seems impactful along with being practical.
 
I read the Alaska/Diomede contributor documentation and repository README, and have been reading the deliberations about the albums idea, and paying particular attention to the arguments people are bringing about preprocessing versus on-the-fly metadata querying. It was also intriguing to observe how the various methods were being evaluated to implement the pipeline and organize DICOM subsets in the most efficient manner.
 
My background mainly includes Python based web development, backend API work, and web based interfaces, so I am excited to work within repositories and explore unfamiliar codebase, while simultaneously building small utilities to understand system better.
 
As I was exploring the repository and the DICOM workflow, my major questions are as follows:

1. Are there any good first issues or small starter tasks in the Diomede repository that I (as a new contributor) can begin with(for example : metadata exploration utilities, small CLI tools, documentation improvement etc)
2. Do you have any suggestions of any publicly available DICOM datasets or sample data sources to which contributors can do local testing?
3. Does the contributors intend to specialise in metadata querying/indexing of local datasets, or do they set out experimenting with integration to viewer pipelines(as an initial point of focus)?
    
   I am looking forward to begin contributing and learn with the community as well as experiment with sample data this week. I would like to thank you that you have provided an open and collaborative environment.
    
   Best regards,
   Aastha Thakkar
   Github : aasthathakkar

[aasthathakkar]

Update(March 23rd, 2026) : I have set up the enviroment and tested pydicom successfully,

-> installed pydicom+matplotlib in venv followed by loading sample CT DICOM files, and printed it's metdata..

Also, visualized the image slice (as attached)

This helped me understand basic metadata reading, which seems useful for the shareable album.

I am a beginner and still learning, but I am ready to do small tasks. Would really appreciate any kind of guidance on what I should try next.

Thank you.

[Pragya-rathal]

Nice progress getting the environment set up and working with sample DICOM files, that’s a good starting point.

If you’re exploring next steps, you might find it useful to:

• look at how metadata fields (like modality, study date, etc.) are structured and normalized
• try writing small filters over metadata (e.g., selecting studies by modality or date)

That connects directly to how albums and querying are being discussed in the repository.

Good luck with the exploration!

[aasthathakkar]

Thanks @Pragya-rathal! I went ahead and implemented (dicom_indexer.py) that recursively scans a local DICOM folder, extracts fields like Modality, StudyDate, StudyInstanceUID, PatientID, SeriesDescription etc. and stores them in a SQLite database via SQLAlchemy. It also skips non-DICOM files gracefully and avoids duplicates using file paths.
Also wrote 13 pytest tests for it using pydicom's own bundled test datasets.
Just opened PR #81 for this — would love any feedback! 🙂