Unravelling a clot-less link between atrial fibrillation and dementia

Assoc Prof James Fisher, Physiology

Introduction

Atrial fibrillation (AF) affects one in four middle-aged adults during their lifetime, and its incidence is expected to double in the next 20 years. AF has an insidious effect on the brain in that people with AF are five times more likely to have a stroke. While cardio-embolic stroke is common in AF, people with AF are more likely to suffer cognitive decline and dementia, even if anticoagulated and in the absence of embolic stroke.

While, the collective evidence suggests that AF-specific mechanisms cause cognitive dysfunction and severe cerebrovascular events, these mechanisms have not been fully elucidated.

The main aim of this research was to investigation of cerebral blood flow regulation in AF, using cutting-edge magnetic resonance imaging (MRI) techniques, and incorporating prospective and interventional studies to provide new mechanistic insights with the potential for significant future clinical impact.

Methodology

Background

Atrial Fibrillation patients and healthy normal sinus rhythm controls underwent brain and cardiac MRI at the Centre for Advanced MRI (CAMRI) at the University of Auckland. All participants performed a visual stimulation task, where they were exposed to a flickering checkerboard stimulus to assess neurovascular coupling responses. In addition, participants performed a CO2 gas challenge, a vasoactive substance, to assess cerebrovascular reactivity. Structural images allowed comprehensive assessment of brain anatomy, and cardiac MRI enabled assessments of cardiac function. Participants also performed cognitive testing at their screening visit using the CANTAB software.

Image processing

Data acquisition for this study has recently concluded, and MRI analysis is currently ongoing. This requires a wide ranging suite of image analysis tools that has been supported by the Centre for eResearch. A series of image preprocessing steps including tissue segmentation of anatomical images, motion correction, susceptibility distortion correction, slice timing correction and estimation of confounds precede the extraction of timeseries data using ROI’s as seen in the figure above. These relate directly to cerebral blood flow changes evoked by our visual stimuli and gas challenge assessments. In addition, the availability of brain imaging software such as FSL allows complex statistical modelling to fully interrogate our data.

Specifically, the Centre for eResearch has facilitated this research by:

Enabling significant file storage, including raw MRI images and post-processing outputs with large file sizes.
Supporting computation on virtual machines to build image processing pipelines.
Supported easy collaboration between researchers to allow seamless data handling.