Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank.
Alfaro-Almagro F, Jenkinson M, Bangerter NK, Andersson JLR, Griffanti L, Douaud G, Sotiropoulos SN, Jbabdi S, Hernandez-Fernandez M, Vallee E, Vidaurre D, Webster M, McCarthy P, Rorden C, Daducci A, Alexander DC, Zhang H, Dragonu I, Matthews PM, Miller KL, Smith SM. · 2018
Read Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank. by Alfaro-Almagro F, Jenkinson M, Bangerter NK, Andersson JLR, Griffanti L, Douaud G, Sotiropoulos SN, Jbabdi S, Hernandez-Fernandez M, Vallee E, Vidaurre D, Webster M, McCarthy P, Rorden C, Daducci A, Alexander DC, Zhang H, Dragonu I, Matthews PM, Miller KL, Smith SM. (2018) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc5770339.txt.
Introduction
Alfaro-Almagro and co-authors describe how the first 10,000 UK Biobank brain scans were processed and checked. UK Biobank is bringing back 100,000 volunteers for brain, heart, and body imaging. The brain work spans six scan types. Raw and processed data from the first group are now open to researchers.
The team built an automated pipeline that turns scans into summary measures. Quality control steps catch faults before release. The paper details the pipeline and the scan protocol behind it. It also reports tests used while the protocol and pipeline took shape. Other groups can use the same pipeline.
Catalog record
| CREATOR | Alfaro-Almagro F, Jenkinson M, Bangerter NK, Andersson JLR, Griffanti L, Douaud G, Sotiropoulos SN, Jbabdi S, Hernandez-Fernandez M, Vallee E, Vidaurre D, Webster M, McCarthy P, Rorden C, Daducci A, Alexander DC, Zhang H, Dragonu I, Matthews PM, Miller KL, Smith SM. |
|---|---|
| YEAR | 2018 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 101 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank By Fidel Alfaro-Almagro, Mark Jenkinson, Neal K Bangerter, Jesper LR Andersson, Ludovica Griffanti, Gwenaëlle Douaud, Stamatios N Sotiropoulos, Saad Jbabdi, Moises Hernandez-Fernandez, Emmanuel Vallee, Diego Vidaurre, Matthew Webster Published in: Neuroimage DOI: 10.1016/j.neuroimage.2017.10.034 ABSTRACT Abstract UK Biobank is a large-scale prospective epidemiological study with all data accessible to researchers worldwide. It is currently in the process of bringing back 100,000 of the original participants for brain, heart and body MRI, carotid ultrasound and low-dose bone/fat x-ray. The brain imaging component covers 6 modalities (T1, T2 FLAIR, susceptibility weighted MRI, Resting fMRI, Task fMRI and Diffusion MRI). Raw and processed data from the first 10,000 imaged subjects has recently been released for general research access. To help convert this data into useful summary information we have developed an automated processing and QC (Quality Control) pipeline that is available for use by other researchers. In this paper we describe the pipeline in detail, following a brief overview of UK Biobank brain imaging and the acquisition protocol. We also describe several quantitative investigations carried out as part of the development of both the imaging protocol and the processing pipeline. Keywords: Epidemiological studies, Image analysis pipeline, Multi-modal data integration, Quality control, Big data imaging, Machine learning INTRODUCTION Biobank project UK Biobank is a prospective cohort study of over 500,000 individuals from across the United Kingdom (Sudlow et al. (2015)). Voluntary participants, aged between 40 and 69 when recruited in 2006–2010, were invited to one of 22 centres across the UK. Blood, urine and saliva samples were collected, samples for genetic analysis and physical measurements were taken, and each individual answered an extensive questionnaire focused on aspects of health and lifestyle. This valuable data resource will provide insight into how the health of the UK population develops over many years and will enable researchers to improve the diagnosis and treatment of common diseases which will inevitably occur in sub-groups of the population. In 2014, UK Biobank began the process of inviting back 100,000 of the original volunteers for brain, heart and body imaging. Imaging data for 10,000 volunteers has already been processed and made available for further research (Biobank, 2016, Biobank, 2017). Due to the wide scope (in population size and number of imaging modalities) of the brain imaging acquisitions, UK Biobank will become an important resource for the research community. The included modalities are T1 (Section T1 pipeline), T2 FLAIR (Section T2 FLAIR pipeline), swMRI (susceptibility-weighted MRI, Section swMRI pipeline), dMRI (diffusion MRI, Section dMRI pipeline), rfMRI (resting-state functional MRI, Section Resting fMRI pipeline) and tfMRI (task fMRI, Section Task fMRI pipeline). It is hoped that ASL (arterial spin labelling, a measure of blood flow in the brain) can be added to the protocol in the near future. Section Acquisition overview briefly describes the acquisition protocol and the acquired modalities. Other large-scale imaging projects include the Maastricht Study with 10,000 subjects (Schram et al. (2014)), the German National Cohort with 30,000 subjects (GNC (2014)) and the Rhineland Study with 30,000 subjects (Breteler et al., 2014). The full potential of UK Biobank stems from its access to NHS (UK National Health Service) records; participants have agreed for their future health outcome information to be linked into the UK Biobank database. Thanks to this, researchers can aim to find novel biomarkers for early diagnosis of different diseases. As described in Sudlow et al. (2015), 1800 of the imaged participants are expected to develop Alzheimer's disease by 2022, rising to 6000 by 2027 (diabetes: 8000 rising to 14,000; stroke: 1800 to 4000; Parkinson's: 1200 to 2800). Greater detail regarding the background to the UK Biobank brain imaging study is given in Miller et al. (2016), a resource paper that also provides an introduction to the types of biological information that can be extracted from the different brain imaging modalities, as well as example results from early analyses from the first 5000 subjects' data, and detailed discussions of the challenges of analysing and interpreting such datasets. In contrast, here we focus more specifically on the methodological advances and implementation details in the image processing and QC pipeline that we have developed for UK Biobank; we describe here version 1.0 of “FBP” (FMRIB's Biobank Pipeline), used to generate the first 10,000 subjects' processed and released data, as of January 2017. We also include several investigations relating to acquisition and analysis pipeline decisions (such as B0 shimming approaches and registration method for dMRI data), with analyses designed to evaluate relative merits of different approaches. Automated pipeline Brain imaging data is not immediately usable for extracting biologically-meaningful information in its raw form. It needs to be processed using a wide variety of software tools, optimised and combined together into a processing pipeline. There are many excellent examples of processing pipelines in the literature. Tools such as those described in Wei et al. (2002) (segmentation), Cook et al. (2006) (diffusion) or Song et al. (2011) (resting-state), are specific solutions for the processing of specific image modalities. However, for the UK Biobank we need a robust pipeline that can process and integrate across many different functional and structural modalities. The minimal preprocessing pipelines developed by the Human Connectome Project (HCP) (Glasser et al., 2013) are good examples of pipelines optimised for the specific needs of a project with very special characteristics. Following the same philosophy, we have developed a fully automated processing pipeline primarily based around FSL software (Jenkinson et al., 2012). In section Automated processing pipelines we describe the pipeline in detail, as well as the choices we have made in its development. Imaging-derived phenotypes and Quality Control A goal of UK Biobank is to identify new relationships between different phenotypic features and human diseases, in the hope that they may be used as biomarkers for early diagnosis. Therefore, in addition to preprocessing the brain imaging data (e.g., aligning modalities and removing artefacts), our pipeline also generates approximately 4350 “imaging-derived phenotypes” (IDPs). These IDPs include summary measures such as subcortical structure volumes, microstructural measures in major tracts (DTI and NODDI measures (Zhang et al., 2012)), and structural/functional connectivity metrics. IDPs for the first 10,000 subjects have been publicly released1 and can be used in combination with other non-imaging data from Biobank to identify disease risk factors and biomarkers. Each IDP is presented as a separate data field within the UK Biobank showcase (except for the rfMRI netmats and node amplitudes, which are saved en masse within “bulk” variables in the database). A detailed description of all IDPs is available in Section S3 of the supplementary material, and the complete list is given online.2 Quality Control (QC) is a very important issue in brain imaging, as poorly executed application (or lack thereof) can compromise the trustworthiness of a study (Bennett and Miller, 2010). This topic has been explored in the literature, although very often research is mostly focused on Quality Assurance (QA) rather than QC. QA is focused on avoiding the occurrence of problems by improving a process while QC is focused on finding possible problems in the output of that process. For example, the Function Biomedical Informatics Research Network (FBIRN) set of recommendations (Glover et al., 2012)