--- tags: TechMon --- # 2025/10/28 (notes) :::info - **Date:** October 28, 2025, 12 pm EST - **Participants:** CM, OE, MN, TS, JHL ::: :::success ## Agenda 1. RMSF Grant outcome 1. TemplateFlow refactor * https://github.com/templateflow/python-client/pull/149 1. Some work on skeleton (https://github.com/nipreps/skeleton/compare/1a507f2...bebda5d) 1. PETPrep - 1. will have a MSc CS student work on https://arxiv.org/pdf/2307.01373 for PETPrep 1. QC participant log of failed/successful runs? (https://github.com/nipreps/petprep/issues/158) 1. PETPrep logo to be added to nipreps website 3. sMRIPrep fast-track/checkpointing ::: ## Notes ### Skeleton * CM will check with RB whether we can use an S3 bucket and we should migrate everything there. * No need for export tree. ### PETPrep * Chat with Yohan and see if we can implement a small (starter) version of his testing and then evaluate how to make it more programmatic/wider for before releases. ### sMRIPrep checkpointing/fast-track * OE to look into AFNI's manifest file. ## Meeting (automated) summary ### Quick recap The team discussed various updates on work commitments, grant applications, and ongoing projects, including a strategic decision to apply for smaller grants and progress on documentation and refactoring efforts. They explored options for organizing test datasets and integration tests for preprocessing pipelines, considering resource constraints and potential improvements to workflow consistency. The group also discussed validation frameworks for upcoming projects, handling cases of missing anatomical data, and improvements to output management, with plans to reconvene in two weeks. ### Next steps * Chris: Check in with Ross about S3 bucket for NiPreps data and migrate all NiPreps data from OSF/Gin to S3 * Chris: Start making NiWorkflows use the new TemplateFlow interface * Oscar: Prepare documentation PRs for TemplateFlow refactor for review * Chris: Push uncommitted skeleton work and create a separate branch on a repository to merge skeleton changes and review in a pull request * Chris: Settle on a way of hosting test datasets as sub-datasets with consistent commands across tools * Martin: Organize a separate meeting to discuss integration testing strategy for PET Prep with Chris, Oscar, and others * Oscar: Finish PET Prep logo SVG conversion and upload to identity repository by end of week * Oscar: Update NiPreps data repository with new DMRI Prep test datasets * Chris: Find and share email from Paul about AFNI manifest files * Oscar: Review AFNI manifest files approach and consider implementing similar system for NiPreps workflows * Oscar: Develop a worked-out example of workflow interface schema definition for feasibility assessment ### Summary #### RSMF The RSMF application was unsuccessful, though the official results have not been published yet. Oscar mentioned he would fill in some topics for discussion and suggested keeping the meeting short given the absences. #### Grant Strategy and Project Updates Oscar discussed the strategic decision to apply for a smaller grant instead of a larger one, explaining that this was a calculated move to receive feedback from a more thorough review process. Chris provided an update on the refactor of the put request, noting that it had been merged but no issues had been reported yet. He mentioned plans to update NiWorkflows to use the new interface and discussed the improved import time for Template Flow. Oscar mentioned initiating PRs for documentation but had not followed up, and Chris shared progress on the skeleton project, which he had been working on across various packages. #### NiPreps Test Dataset Organization Oscar and Chris discussed the organization of test datasets for NiPreps, considering options like using a "super dataset" or individual sub-datasets. Chris mentioned working on standardizing the hosting of test datasets and agreed to check with Ross about using an S3 bucket for NiPreps data. They decided against using an export tree structure for S3, preferring a simpler key-based approach. The conversation concluded with a brief mention of moving on to discuss pet prep. #### fMRI and PET Integration Tests The team discussed integration tests for fMRI and PET preprocessing pipelines. Chris explained that Johan's tests were initially a research project, but they are now considering using them as integration tests. They discussed the possibility of running these tests before each release, but Chris expressed concerns about running them in CI due to resource constraints and the potential for inconsistent results. Martin suggested decomposing the tests to focus on specific preprocessing steps like anatomical normalization. The team agreed to have a separate meeting to discuss this topic in more detail, as they want to think ahead about incorporating these tests into their workflow. Oscar suggested starting with a validation effort for the first version of PET Prep, similar to what they did for fMRI Prep. #### Validation Framework for Pre-Project Oscar and Martin discussed a validation framework for a pre-project that will run for 4 months starting in February, similar to their previous work on fMR Prep. They agreed to set up a validation framework and not cut the LTS until a certain threshold of improvement is reached. Taylor shared his experience with ASL prep validation, which involves reviewing HTML reports and figures for quality control. The group discussed the need for a more standardized and affordable validation process for new LTS releases. They also touched on a recent issue with PET Prep failing on some subjects due to lack of anatomical data, and considered options for handling such failures in the workflow. #### Improving Missing Scan Notifications The team discussed handling cases where anatomical scans (T1-weighted MRI) are missing from datasets. Martin expressed interest in improving how users are informed about failed processing due to missing scans, with Oscar suggesting that fMRI Prep currently stops processing entirely if any anatomical data is missing. Chris proposed creating a participants.tsv file to track successful and failed runs, while Oscar shared improvements made to MRQC logging, including colored output and structured formatting, which could be useful for fMRI Prep. #### Website Updates and Data Integration Martin requested Oscar to add a Pet Prep logo to the NiPreps website, and Oscar confirmed he was working on it, aiming to complete it by the end of the week. Oscar also discussed his progress on integrating DMRA prep with FMR prep, noting that the SMRI Prep connection was working but some datasets were failing to download. He planned to update the NiPreps data repository and mentioned the need to rely on Nightpipe for checking tasks. fMRI Workflow Output Management Discussion The team discussed the handling of derivatives and output directories in fMRI Prep and SMRI Prep workflows. Chris explained how derivatives are managed and the default output structure, while Oscar suggested a potential approach to integrate SMRI Prep outputs into the workflow. Taylor shared current practices for handling fMRI and ASL data separately. The group explored ideas for improving workflow consistency, including the use of manifest files and schema definitions. Chris mentioned AFNI's approach to output management, which Oscar agreed to investigate further. The conversation ended with plans to reconvene in two weeks.