gantt
title Semester Project Gantt Diagram
dateFormat DD.MM.YYYY
axisFormat %d.%m
section Background
Literature review (DWI, GQI, MRI) :done, research, 24.11.2025, 2d
Study Nipreps book (DWI preprocessing) :done, nipreps, 24.11.2025, 3d
section Development
Dev environment setup :done, devenv, 25.11.2025, 1d
Study GQI codebase :done, codebase, 25.11.2025, 3d
Create baseline prediction tests :done, simple_tests, 26.11.2025, 2d
Develop Marimo validation notebooks :done, notebooks, 26.11.2025, 2d
Implement roundtrip voxel prediction test :done, cross_tests, 01.12.2025, 2d
Create unseen data prediction notebook :done, unseen_data, 03.12.2025, 4d
Add the GQI predict example to dipy doc :done, predict_example,04.12.2025, 11.12.2025
Create prediction_kernel tests :done, prediction_kernel ,07.12.2025, 2d
Double check the existing tests :done, dc_test, 09.12.2025, 1d
Double check the existing notebooks :done, dc_notebooks, 09.12.2025, 1d
Coordinate with DIPY experts :done, coor, 10.12.2025, 12.12.2025
Fine tune the code for the DIPY PR based on feedback :done, prpr, 12.12.2025, 17.12.2025
section Milestones & <br>Deliverables
Prof. Esteban – Kickoff meeting :done, milestone, 25.11.2025, 0d
Prof. Esteban – Progress check-in :done, milestone, 27.11.2025, 0d
Create Gantt diagram :done, gantt, 28.11.2025, 1d
Design impact slide :done, impact, 28.11.2025, 1d
Draft report outline :done, r_draft, 01.12.2025, 1d
Prof. Esteban - Second progress check-in:done, milestone, 03.12.2025, 0d
Create presentation DWI slide :done, p_draft, 06.12.2025, 09.12.2025
Prof. Esteban - Third progress check-in :done, milestone, 09.12.2025, 0d
PR on Prof. Esteban repository :done, milestone, 10.12.2025, 0d
PR on DIPY repository :done, milestone, 11.12.2025, 0d
Finish report :done, f_rep, 11.12.2025, 16.12.2025
Finish presentation :done, f_pres, 11.12.2025, 18.12.2025
Prof. Esteban - Final progress check-in :done, milestone, 15.12.2025, 0d
Code and documentation handout :done,crit, milestone, 16.12.2025, 0d
Final presentation :done, crit, milestone, 18.12.2025, 0d
303.1 Semester Project:
Extending DIPY’s Generalized Q-ball Imaging model to make predictions
Project Overview
This project is part of the 303.1 semester project course.
The goal is to contribute to the DIPY framework by extending the Generalized Q-ball Imaging (GQI) model with predictive capabilities. The key objectives are:
- Implementing a
predict()function to theGeneralizedQSamplingFitclass - Thoroughly documenting the new code
- Providing comprehensive and meaningful tests
- Getting the prediction pull request accepted by DIPY
The project has progressed to the point where a pull request has been submitted to the DIPY repository for integrating the new predict() function. This pull request is currently under review by the DIPY development team, and feedback is being incorporated to ensure it meets the library’s standards and requirements.
Project Timeline
The following Gantt diagram provides a visual overview of the project schedule, highlighting completed tasks, active work, and upcoming milestones. It tracks progress from initial research and development through testing, documentation, and final submissions.
```
Key Achievements
The project has successfully delivered a complete, production-ready implementation with validated performance:
Mathematical Implementation: Successfully implemented the
predict()function for theGeneralizedQSamplingFitclass using a Tikhonov-regularized Moore-Penrose pseudo-inverse approach, along with the essentialprediction_kernel()helper function that ensures numerical stability through regularization.Comprehensive Testing Suite: Developed rigorous tests covering:
- Unit tests (kernel properties, pseudo-inverse validation),
- API tests (single/multi-voxel support)
- Integration tests (roundtrip accuracy).
Validated Roundtrip Tests Performance: Achieved high roundtrip accuracy with \(r = 0.9744\) single-voxel correlation and \(r = 0.9800\) mean multi-voxel correlation on the
small_101Ddataset1.Robust Generalization: Demonstrated strong generalization to unseen gradient configurations using the
taiwan_ntu_dsidataset2, with local correlation maps showing promising performance across anatomical regions and successful capture of white matter tract structures.Educational Resources: Created complete documentation pipeline including a DIPY tutorial notebook and project-specific validation notebooks demonstrating the different obtained results.
Community Integration: Submitted a pull request to the DIPY repository following established contribution standards with active coordination with DIPY maintainers for review and integration.
Project Documents And Resources
Access the key deliverables and supplementary materials for this project:
- Implementation Report: Comprehensive technical documentation covering the
predict()function implementation, mathematical foundations, testing methodology, results analysis, and extra resources for further exploration. - Presentation Slides: Slide deck summarizing the project objectives, methodology, achievements, and more.
- Daily Journal: Chronological log of development activities, challenges encountered, and progress updates throughout the project timeline.
- Sources and References: Bibliography, citations, and additional reading materials used in this project.
For results examples and code demonstrations, refer to the notebooks and examples linked in the Implementation Report.