SINAPSE has substantial expertise in a range of imaging techniques including CT and MR brain imaging, CT angiography, MR angiography, intra-arterial angiography, PET, SPECT and multi-modality imaging. Well-validated and extensively tested, structured, standardised assessment tools are in place for all of these applications and can readily be adapted to other areas.
Across the SINAPSE collaboration, we have a wide range of image processing capabilities, ranging from subjective radiological interpretation using standardised validated imaging reading formats, through to complex image analysis software tools for processing tractography or perfusion imaging data. While many of these are under active development or in use in ongoing projects, we are actively engaged in examining which programmes might be made more widely available and have already made available several standardised templates for radiological interpretation.
We have developed an image analysis toolbox which includes the brain volume loss assessment tool for older subjects, a set of standard scans for rating white matter lesions and instructions about a range of white matter lesion scales, and a validated method for rating brain microbleeds. These are all available through the image analysis toolbox section of the SFC Brain Imaging Research Centre website.
Further tools that are under development and potentially available on a collaborative basis include: brain volume measurement and segmentation tools; white matter lesion volume measurement; tractography processing methods; perfusion image analysis; blood-brain barrier permeability analysis; spectroscopy (both single voxel, chemical shift imaging and brain temperature assessment); and magnetisation transfer measurements. We also use image registration tools and voxel-based image analysis techniques.
SINAPSE can utilise standard packages such as SPM and FSL for functional and structural image analysis, and DTIstudio and 3Dslicer for diffusion tensor imaging data analysis. In addition, our researchers are experienced in writing custom software in Matlab, C++ and Python to perform other data analysis functions or to streamline multiple image processing tasks.
Due to the number of image processing techniques available, the brain's anatomical structure and functional activations can be depicted with several different methods such 2D sections, interactive 3D models, flattened brains, glass brains, etc. These various image representations are used in different contexts for different purposes. While clinicians seek to identify altered anatomical structures, medical students need to learn about neuroanatomy and cognitive researchers attempt to localise functional activations. A widely neglected issue across all of these domains is the effect of data visualisation styles and individual skills on the assessment of brain images. Research in Dundee is being undertaken to explore how different image formats support reasoning skills in neuroimaging. The aim is to propose guidelines for user-friendly image rendering, and develop training recommendations for interpreting brain images.
Aberdeen researchers use a number of imaging and data analysis techniques, such as fMRI, optimised voxel-based morphometry (VBM), diffusion tensor imaging with tractography, and kurtosis imaging, as well as T2 relaxography imaging, to investigate structural and functional correlates of development and ageing. For example VBM has been used to demonstrate differences in brain structure in autism, which underlie functional differences demonstrated on fMRI. In ageing, VBM has been used to demonstrate retention of white matter volume in those maintaining intelligence in late life.
Brain MRI research in the Aberdeen Birth Cohorts of 1921 and 1936 is centred on life-long cognitive change, cerebral reserve, predictors of successful cognitive ageing and causes of white matter hyperintensities. Data held includes volumetric, fMRI, diffusion tensor imaging, white matter analyses and measures of entropy.
Researchers at the University of Edinburgh have developed an algorithm called MCMXXXVI (1936). It is a method for segmenting tissues and performing volume analysis in MRI. It has been developed to segment tissues and abnormalities, such as white matter lesions, automatically and with high accuracy. It is being used to understand how structural changes in the brain contribute to cognitive decline in normal ageing using multimodality MRI, and involves more than 1,000 older healthy subjects from the Lothian Birth Cohort 1936 whose cognitive ability was originally studied at age 11 years, and is being tested again at age 70 to 71.
One of our key aims is to ensure that researchers from different institutions utilise standardised procedures for processing image data, to ensure that results from different centres can be compared. In this way, we aim to be leaders in multi-centre neuro imaging research.
We can ensure that images collected in one centre are the same as those collected elsewhere. To measure any scanner related differences, a project called Calibrain examined the differences in brain structure and function of the same volunteers imaged on 3 scanners in 3 different centres. Results from the Calibrain project demonstrate that when multicentre brain MRI is carried out in Scotland, the data collected in one centre is comparable to that in another centre.
