Optimizing automated white matter hyperintensity segmentation in individuals with stroke

Biomedical imaging is a driver of scientific discovery and a core component of medical care and is being stimulated by the field of deep learning. While semantic segmentation algorithms enable image analysis and quantification in many applications, the design of respective specialized solutions is non-trivial and highly dependent on dataset properties and hardware conditions. We developed nnU-Net, a deep learning-based segmentation method that automatically configures itself, including preprocessing, network architecture, training and post-processing for any new task. The key design choices in this process are modeled as a set of fixed parameters, interdependent rules and empirical decisions. Without manual intervention, nnU-Net surpasses most existing approaches, including highly specialized solutions on 23 public datasets used in international biomedical segmentation competitions. We make nnU-Net publicly available as an out-of-the-box tool, rendering state-of-the-art segmentation accessible to a broad audience by requiring neither expert knowledge nor computing resources beyond standard network training.

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.

Objectives To review the evidence for an association of white matter hyperintensities with risk of stroke, cognitive decline, dementia, and death. Design Systematic review and meta-analysis. Data sources PubMed from 1966 to 23 November 2009. Study selection Prospective longitudinal studies that used magnetic resonance imaging and assessed the impact of white matter hyperintensities on risk of incident stroke, cognitive decline, dementia, and death, and, for the meta-analysis, studies that provided risk estimates for a categorical measure of white matter hyperintensities, assessing the impact of these lesions on risk of stroke, dementia, and death. Data extraction Population studied, duration of follow-up, method used to measure white matter hyperintensities, definition of the outcome, and measure of the association of white matter hyperintensities with the outcome. Data synthesis 46 longitudinal studies evaluated the association of white matter hyperintensities with risk of stroke (n=12), cognitive decline (n=19), dementia (n=17), and death (n=10). 22 studies could be included in a meta-analysis (nine of stroke, nine of dementia, eight of death). White matter hyperintensities were associated with an increased risk of stroke (hazard ratio 3.3, 95% confidence interval 2.6 to 4.4), dementia (1.9, 1.3 to 2.8), and death (2.0, 1.6 to 2.7). An association of white matter hyperintensities with a faster decline in global cognitive performance, executive function, and processing speed was also suggested. Conclusion White matter hyperintensities predict an increased risk of stroke, dementia, and death. Therefore white matter hyperintensities indicate an increased risk of cerebrovascular events when identified as part of diagnostic investigations, and support their use as an intermediate marker in a research setting. Their discovery should prompt detailed screening for risk factors of stroke and dementia.