Have your say in the design of BIP - The UK Biobank Brain Imaging Pipeline!

2260 

Abstract Submission 

Fidel Alfaro Almagro¹, Stephen Smith¹

¹University of Oxford, Oxford, Oxfordshire

Fidel Alfaro Almagro  
University of Oxford
Oxford, Oxfordshire

Stephen Smith  
University of Oxford
Oxford, Oxfordshire

The image processing pipeline for UK Biobank brain imaging (1), has been used to process more than 72k UKB datasets, and >500 papers have used its outputs. The pipeline works with T1, T2 FLAIR, swMRI, dMRI, rfMRI, tfMRI (and ASL in recent subjects) and consists of a mix of scripts in bash, Python and Matlab, with the underlying tools coming primarily from FSL and FreeSurfer (Fig 1). It applies image processing and QC, and generates thousands of Imaging-Derived Phenotypes (IDPs). The pipeline was optimised for UKB data and has been adapted for other studies with some effort (2, 3, 4).

By 2023, 72k brain imaging datasets had been acquired and processed. 63k were usable data from the first imaging visit and 5k from the second. The goal is 100k first-visit and 60k repeat-visit scans. Fig 2 shows pairwise associations between 4k brain IDPs and 27k non-imaging variables.

The pipeline has maintained almost perfect backwards compatibility so that all outputs (for new scans) are compatible with existing outputs for previous subjects/scans. Therefore, changes have been limited to adding new processing (e.g., the recent addition of QSM processing). However, maintaining backwards compatibility limits the ability to incorporate improved algorithms and software, and we are now planning a complete rewrite of the pipeline, named BIP (Brain Imaging Pipeline).

Improvements being considered include:

Pipeline framework: fully Python-based and using three main components:
- Fslpy (5): Python library to perform FSL calls through wrappers
- File-Tree (6): Python library to abstract out filenames and folder structure
- FSL-pipe (7): Easy (to use and scale) declarative pipelining Python library

Usage
- The pipeline should be easy to configure (with as few configuration files as possible) for datasets with configurations different from UKB (but not too different)

Registration
- Multi-modal (T1, T2-FLAIR, dMRI) volumetric registration using MMORF and multi-modal UKB-derived OMM templates (8,9)
- Surface registration using MSMAll (10)

Surface ("CIFTI" HCP) processing (note that this has now been added to the pipeline)
- fMRI projected onto the surface

fMRI
- PROFUMO as a complement/replacement for group-ICA+dual-regression (11)

Diffusion
- Better cleaning (Gibbs ringing correction, modelling Rician noise)
- Improved Eddy-current correction (within-volume motion, long-time eddy currents)
- Replace autoPtx with XTRACT (reference)
- Improved NODDI model

QC
- Fully automated (if possible), improved QC

We want to ask the neuroimaging community for their opinion on our plans and what else we can do with UKB data in BIP, starting with a set of surveys. We encourage suggestions for solidly established complementary/improved methods. We will engage experts in each area, with expert working groups to evaluate suggestions, including considering these criteria:
- Feasible: Given the amount of data to process, the tools must be easy to implement (and free to use) and not too memory/disk/processing intensive
- Testable: The method should be easily testable against similar methods with clear and measurable criteria to establish which is best for UKB (and other) data.
- Fit for framework: Once the framework has been decided, adding pieces that do not fit well with the framework would make the pipeline too complex.
- Aligned with UKB data and goals: We will consider suggestions sufficiently close to the UKB brain imaging protocol.

UKB is an epidemiological study that needs a robust and fully automated pipeline to process all its data. Therefore, once BIP is considered "finalised", backwards compatibility must be maintained. Thus, if you want to have your say on this processing pipeline, the moment to do that is now!

Methods Development

Databasing and Data Sharing

Workflows ¹

Informatics Other ²

Data analysis

Design and Analysis

Informatics

MRI

Open-Source Code

Open-Source Software

Workflows

Other - UK Biobank