NeuroImaging Laboratory

The Brain and Ageing Research Program's NeuroImaging Laboratory was established in 1991 in response to a need for the assessment of brain images. This Laboratory brings together the experience of a diverse group of researchers whose interdisciplinary work combines clinical and research psychiatry, neurosciences, and the theoretical advances in mathematics,statistics, information theory, and engineering.

Great advances have been made in the acquisition of brain image data e.g., the development of new and powerful scanners and imaging protocols for both structural and functional brain magnetic resonance imaging (MRI). Investigations into brain structure and function require a diverse range of tools to analyse and visualize the images acquired. Researchers at the Laboratory are developing new algorithms and methods, as well as using existing software packages, in their research. The Laboratory houses a high performance cluster (HPC) computer, 12 dedicated workstations (both Windows and Linux) and two data archival systems of about 30 terabytes. The MR images studied include 3D T1-weighted scans, T2-weighted (such as FLAIR sequence) scans, DTI (diffusion tensor imaging), 1H MRS (magnetic resonance spectroscopy), task functional MRI (fMRI), resting-state fMRI, Gd-perfusion MRI (pMRI), susceptibility weighted imaging (SWI) and will extend to ASL (arterial-spin labelling) in the near future. Our Laboratory has used both PET and SPECT scans in the past, and continues to use these, and other, imaging modalities in current studies.

The NeuroImaging Laboratory collaborates with universities within Australia and overseas, including Australian National University, Canberra Australia; The University of Newcastle, Newcastle, Australia; Johns Hopkins University, Baltimore, USA; Northwestern University, Chicago, USA; Simon Fraser University, Vancouver, Canada; Beijing Normal University, Beijing, China and Shanghai Jiao Tong University, Shanghai, China. The Laboratory is also part of world wide studies, such as ENIGMA, a network in imaging geonomics.

Research opportunities

We welcome both postgraduate (Master by Research & PhD programs) and undergraduate (Honours program) students to carry out research projects using our Neuroimaging Laboratory data and facilities (this includes excellent computing resources).

Potential applicants should contact Dr Wei Wen in the first instance.

Research Areas

Postgraduate projects: For details of available projects, please contact Dr Wei Wen.

Undergraduate projects: (Faculty of Medicine Honours Program)

Structural Brain Networks

Traditionally, the study of the relationship between brain structure, brain ageing and associated neuropsychiatric disorders, has been to examine circumscribed atrophy or other abnormalities of various grey, or white matter regions of interest, by using 3D T1-weighted structural MRI scans. However, there are limitations with this approach. One such limitation is that in normal ageing, and in brain disease such as Alzheimer's Disease or schizophrenia, a large number of structures show atrophy or abnormality; making the significance of any one structural change difficult to establish. Furthermore, this approach does not take into account the fact that the brain functions as a network of inter-connected regions, and it is the abnormality in the network that is more indicative of functional impairment [1]. The emerging, and rapidly moving, network approach, based on graph theory has the significant advantage of a rich, structural description, which allows efficient computation and comparison of different connection topologies within a common theoretical framework. Structural brain networks images are constructed using either structural brain scans, such as DTI, or 3D T1-weighted scans. The NeuroImaging Laboratory carries out structural brain network research using both of these type of scans.

DTI tractography based structural brain network

This type of structural brain network research is based on the diffusion tensor imaging, fibre tract derived, connectivity between grey matter regions, presenting anatomical connections between brain regions. Recent advances in DTI and tractography have facilitated the non-invasive mapping of structural networks in the human brain, at an individual level i.e., a graph can be constructed for each brain. In DTI-based tractography, what is reconstructed does not represent actual axonal fibres, which have a diameter of only a few microns, where the voxel size of DTI scans is usually more than 2mm³. Nevertheless, the DTI scan does reflect the macroscopic configuration of "axonal bundles", which are large enough to be visually appreciated.


Both T1-weighted structural MRI and diffusion tensor imaging (DTI) are acquired for each individual being assessed as part of a particular study. Grey matter (GM), white matter (WM) and cerebrospinal fluid (CSF) are segmented then a boundary of GM and WM is parcellated into individual cortical areas (66 in fig 1), and then are mapped into DTI space. The raw dDTI is processed to obtain fibre-tractography for calculating the connection between cortical regions and the number of fibres that connect them. A binary and a weighted matrix are established. A graph (network) is then constructed using the information from the connectivity matrices. The topological properties of the network(s) are then computed and compared with those of other participants in the study,

BARP NeuroImaging Laboratory Study:
In the NeuroImaging Laboratory, we have successfully carried out a DTI network study using graph theory based analyses of DTI tract-derived connectivity, using a sample of 342 healthy individuals aged 72-92 years [2]. The cognitive domains assessed included: processing speed, memory, language, visuospatial and executive functions. We examined theses cognitive assessments with both the connectivity of the whole brain network, and individual cortical regions. We found that the efficiency of the whole brain network of cortical fibre connections had an influence on processing speed, visuospatial and executive functions. Correlations between connectivity of specific regions and cognitive assessments were also observed, e.g. stronger connectivity in the regions such as superior frontal gyrus and posterior cingulate cortex were associated with better executive function. Similarly, the relationship between regional connectivity efficiency, age and greater processing speed were significantly correlated with better connectivity of nearly all cortical regions. For the first time, regional anatomical connectivity maps related to processing speed, visuospatial and executive functions in the elderly, were identified.

DTI tractography based structural brain network

This type of network uses brain regions as a set of nodes and the interregional correlations of regional cortical thickness, or cortical and subcortical grey matter volumes, across individual brains as edges between the nodes. This type of structural brain network is constructed using 3D T-1 weighted structural (sMRI) scans, and because it uses interregional correlations of grey matter thickness, or volume of brain regions, only one graph can be constructed for a group of subjects. The notion of morphological correlations has been widely used to study correlated evolution in mammalian brain structure, or to infer structural connectivity between human brain regions.

Related BARP NeuroImaging Laboratory Study:
Recently, one of the brain connectivity projects, run out of the NeuroImagining Lab, examined normal ageing from the perspective of topological patterns of structural brain networks, constructed from 2 groups of healthy individuals, with an age difference of approx. 20 years [3]. Based on graph theory, we constructed structural brain networks using 90 cortical and subcortical regions as a set of nodes, and the interregional correlations of grey matter volumes across individual brains as edges between nodes; and further analysed the topological properties of the age-specific networks. We found that the structural brain networks of both cohorts had small-world architecture, and the older cohort (N=374; mean age=66.6 years, range 64-68 years) had lower global efficiency but higher local clustering in the structural brain networks compared with the younger cohort (N=428; mean age=46.7 years, range 44-48 years). The older group had reduced hemispheric asymmetry and lower centrality of certain brain regions, such as the bilateral hippocampus, bilateral insula, left posterior cingulate and right Heschl gyrus, but that of the prefrontal cortex (PFC) was not different. These structural network differences may provide the basis for changes in functional connectivity, and indeed, cognitive function as we grow older.

Related Publications

[1] Wen W, He Y, Sachdev P. Structural brain networks and neuropsychiatric disorders. Current Opinion in Psychiatry 2011a: 24: 219-225.
[2] Wen W, Zhu W, He Y, Kochan NA, Reppermund S, Slavin MJ, Brodaty H, Crawford J, Xia A, Sachdev P. Discrete neuroanatomical networks are associated with specific cognitive abilities in old age. Journal of Neuroscience 2011b: 34: 1204-1212.
[3] Zhu W, Wen W, He Y, Xia A, Anstey KJ, Sachdev P. Changing topological patterns in normal aging using large-scale structural networks. Neurobiology of Aging (In Press), doi:10.1016/j.neurobiolaging.2010.06.022

Diagnosis and prediction of mild cognitive impairment and Alzheimer's disease using pattern recognition and methods

Early and accurate diagnosis,and prediction, of mild cognitive impairment (MCI) and Alzheimer's disease (AD) is particularly important in facilitating the development of treatments which may prevent AD, or slow its progression. However, this is particularly challenging due to the subtlety of brain changes at the very early stages of the disease.

Pattern recognition, is essentially, the study of how machines can observe the environment, learn to distinguish the patterns of interest , and make decisions about the observed patterns. The study of pattern recognition is an interdisciplinary field combining areas such as mathematics, statistics, engineering, computer science, psychology and physiology. Human beings are particularly apt at recognising patterns, from an early age they are able to recognise a face, spoken words and written texts. Extensive research has been conducted to enable machines to match human perception, or in some cases, to exceed human ability. Pattern recognition methods have attracted much attention in brain imaging and computational neurosciences. Pattern recognition approaches enable automated exploration of multivariate relationships among high-dimensional features to differentiate between groups, the construction of non-linear models and automated classification, of scans of study participants, on an individual basis. In addition, multimodal techniques are playing an increasingly important role in characterising structural and functional profiles, in both normal and diseased brains.


Related BARP NeuroImaging Laboratory Studies

In studies conducted in the NeuroImaging Laboratory, pattern recognition algorithms have been used for the diagnosis and prognosis of MCI in 2 aspects: 1) automated detection of amnestic MCI in community-dwelling elderly individuals, and 2) predicting the cognitively normal individuals at increased risk of developing MCI. Samples of these 2 studies were drawn from the Sydney Memory and Ageing Study. The aim of the first study, was to achieve automated detection of amnestic MCI using a combination of spatial atrophy and white matter alterations of the brain [1,2]. The second study predicted the transition from normal cognition to MCI, using neuropsychological assessments and structural morphological measures; which is, to the best of our knowledge, the first pattern recognition study to do this. Our pattern recognition methods achieved high level of performance with both these studies. In addition, the performance was greater when using a combination of multiple modalities of measures than when using either of these methods alone. Further to this, discriminating features identified in the schema enhance the understanding of AD progression. Our pattern recognition schema offers the potential to facilitate the early detection of MCI, and the early start of interventions designed to prevent or slow the development of AD and other dementias.

Related Publications

[1] Cui Y, Wen W, Lipnicki DM, Beg MF, Jin JS, Luo S, Zhu W, Kochan NA, Reppermund S, Zhuang L, Raamana PR, Liu T, Trollor JN, Wang L, Brodaty H, Sachdev PS. Automated detection of amnestic mild cognitive impairment in community-dwelling elderly adults: A combined spatial atrophy and white matter alternation approach. NeuroImage (In Press).
[2] Cui Y, Lui B, Luo S, Zhen X, Fan M, Liu T, Park M, Jiang T, Jin JS. Identification of conversion from mild cognitive impairment to Alzheimer's disease using multivariate predictors. PLoS One, 2011a: 6:e21896.

White matter integrity in mild cognitive impairment

Cortical atrophy in the medial temporal lobe has been widely accepted as a pathological hallmark of Alzheimer's disease (AD). However, convergent evidence from post-mortem brain studies, and animal models suggest that AD could also be a disconnection syndrome independent of cortical grey matter atrophy. Since the advent of diffusion tensor imaging (DTI), neuroimaging studies have attempted to capture in-vivo white matter microstructural changes in AD; although, white matter changes in mild cognitive impairment (MCI), a preclinical stage of AD is less known.

Related BARP NeuroImaging Laboratory Studies

In a project relating to white matter integrity and its role in MCI, we aimed to 1) investigate spatial patterns of in-vivo white matter alterations in MCI and its subtypes; 2) examine whether white matter changes are more sensitive to grey matter atrophy in the early detection and diagnosis of AD; and 3) examine neural correlates of episodic memory loss in early AD [1, 2, 3].

By using whole brain DTI analysis method, tract-based statistical analysis [part of FMRIB's Software Library (FSL) program, Oxford Centre for Functional Magnetic Resonance Imaging of the Brain, Oxford University, UK), we found significant reduction of white matter integrity in the frontal, temporal, parietal and occipital regions, together with several commissural, association and projection fibres in amnestic subtype of MCI (aMCI) subjects. In contrast, non-amnestic subtype of MCI (naMCI) showed more widespread white matter damage, but without the involvement of the medial temporal lobe white matter. These results suggest that aMCI is best characterised by pathology consistent with early Alzheimer disease, whereas underlying pathology in naMCI is more hetreogeneous. Since whole brain DTI analysis has limitations in studying specific white matter tracts; including the fornix, parahippocampal cingulum, uncinate fasciculus, association fibre (including inferior longitudinal fasciculus and one projection fibre) and the corticospinal tract, to further investigate which white matter tract underlies episodic memory loss in the preclinical stages of dementia.

Related Publications

[1] Chua TC, Wen W, Chen X, Kochan N, Slavin MJ, Trollor JN, Brodaty H, Sachdev PS. Diffusion tensor imaging of the posterior cingulate is a useful biomarker of mild cognitive impairment. American Journal of Geriatric Psychiatry, 2009: 17: 602-613
[2] Chua TC, Wen W, Slavin MJ, Sachdev PS. Diffusion tensor imaging in mild cognitive impairment and Alzheimer's disease: A review. Curent Opinion in Neurology, 2008: 21:83-92
[3] Zhuang L, Wen W, Zhu W, Trollor J, Kochan N, Crawford J, Reppermund S, Brodaty H, Sachdev PS. White matter integrity in mild cognitive impairment: a tract-based spatial statistics study. Neuroimage, 2010. 53:16-25

Morphology of cortical surface: cortical folding pattern and sulcal width

The cortical folding process begins very early, starting from 10 weeks of foetal life. Therefore, perturbations of cortical development can provide us with important clues in the resulting morphology. Such morphological changes in the brain are associated with ageing, and this is possibly related to the thinning of the gyri, due to reduction in gyral grey matter and white matter. Sulcal widening is commonly used by radiologists as a measure of cortical atrophy in the clinical setting.

Related BARP NeuroImaging Laboratory Studies

Using tools developed at the Laboratory and in public domain pipelines (including BrainVisa, FreeSurfer, FSL and SPM), we have examined the brain's structures both using voxel-based cortical morphometry analysis, and surfaced-based cortical morphometry analysis. The imaging measures that were used included cortical density, cortical thickness, cortical sulcal span, and cortical sulcal index. Sulcal index reflects the complexity of sulcal folding. We also computed the region interrelationships to examine the differences between 2 groups divided by cognitve function in the elderly [1,2].




Fig 1: Thickness and sulcus measurements by surface-based cortical morphometry analysis.

Fig 2: Region interrelationships measurements by voxel-based cortical morphometry analysis

Related Publications

[1] Liu T, Wen W, Zhu W, Kochan NA, Trollor JN, Reppermund S, Jin JS, Luo S, Brodaty H, Sachdev PS. The relationship between cortical sulcal variability and cognitive performance in the elderly. NeuroImage 2011. 56:865-873.

[2] Liu T, Wen W, Zhu W, Trollor J, Reppermund S, Crawford J, Jin JS, Luo S, Brodaty H, Sachdev P. The effects of age and sex on cortical sulci in the elderly. NeuroImage 2010. 51:19-27.