Browsing by Author "Robertson, Frances"

Now showing 1 - 9 of 9
  • Thumbnail Image
    Item
    Open Access
    Brain morphometry of HIV-infected children on early antiretroviral therapy (ART) from age 5 to 9 years
    (2020) Nwosu, Emmanual Chukwubuikem; Robertson, Frances; Meintjies Ernesta
    As of 2017, 1.8 – 2.1 million children vertically infected with HIV were living in sub-Saharan Africa, of whom an estimated 320, 000 were in South Africa. Since implementation of the prevention of mother to child transmission (PMTCT) strategy, the infection rate has reduced substantially. More recently, the World Health Organisation's (WHO) recommendation of early antiretroviral therapy (ART) initiation for children with perinatal HIV infection has considerably decreased the immediate effects of perinatal HIV infection, including mortality and morbidity. Despite this, not much is known about the long-term outcome of continued ART on early-treated, perinatally HIV-infected children. Early HIV invasion of the developing brain is associated with neurodevelopmental delays and neurocognitive deficits including encephalopathy, slower processing speed, language impairment, lack of concentration and attentiveness, and psychomotor slowing. Alterations in the neurodevelopmental trajectories of brain morphology, including cortical thickness and folding (gyrification) and sub-cortical volumes may be related to the observed neurocognitive deficits during a critical period of brain development spanning from mid-childhood into early adolescence (age 5 -13 years). The effects may be studied using structural magnetic resonance imaging (MRI) and automated segmentation software. FreeSurfer (https://surfer.nmr.mgh.harvard.edu/) is a valuable tool for investigating brain morphology but was not originally designed for segmenting pediatric brains. In this study we therefore first validate the latest FreeSurfer version 6.0.0 against manual segmentation for the study of pediatric HIV. We then assessed the long-term effects of perinatal HIV infection, early ART initiation as well as clinically designed ART interruption, HIV-related encephalopathy, disease severity at ART initiation and immune health measures on the developmental trajectories of cortical thickness and folding (gyrification) over the period from 5-9 years. Study participants were 141 children (75 HIV+, 66 uninfected controls; 72 male) from the Cape Town arm of the children with HIV early antiretroviral therapy (CHER) clinical trial. HIV+ children were randomized at age 6 -12 weeks to receive either immediate limited ART for 40 or 96 weeks, to be restarted when clinical and/or immunological criteria were met, or to start ART only when they developed HIV symptoms or CD4 percentage dropped below 20% (25% in the first year) as per guidelines at the time. Uninfected controls comprised children born to HIV+ mothers (HIV-exposed uninfected (HEU)) or uninfected mothers (HIV-unexposed (HU)) and were recruited from an interlinking vaccine trial. MRI scans were performed at time points around their 5th, 7th and 9th birthdays, in accordance with protocols approved by the human research ethics committees of the Universities of Stellenbosch and Cape Town and voluntary informed consent was received from either participants or their guardians. Both automated and manual methods were used to segment brain regions from high-resolution structural MRI scans. In addition, FreeSurfer was used to examine cross-sectional differences in cortical thickness and gyrification over the cortical surface at age 5. Linear mixed-effects models were used in conjunction with FreeSurfer's longitudinal processing stream to calculate and compare the annual rate of change in cortical thickness and gyrification between ages 5 and 9 in HIV+ children and controls. Results showed that automated FreeSurfer segmentation tended to overestimate volumes of all structures relative to manual segmentation, except the left caudate nucleus. Consistency and agreement between methods were highest for the putamen (Consistency: right ICC=0.89, left ICC=0.90; agreement: right ICC=0.84, left ICC=0.83) and lowest for the corpus callosum (consistency ICC=0.64, agreement ICC=0.26). There were no subcortical volume differences between HIV+ children and controls, except the globus pallidus which was smaller in HIV+ children using both manual and automated segmentation. Subsequent cross-sectional FreeSurfer analyses showed widespread regional increases in cortical thickness and decreases in gyrification at age 5 years, related to the effects of perinatal HIV-infection and early ART initiation. Clinically designed interruption led to thicker cortex in the left rostral middle frontal and right insula regions and, lower left precuneus and right superior frontal, as well as higher lateral occipital gyrification compared to HIV- controls. There were significant regional differences due to HIV severity based on CDC classification and viral burden at enrolment both in cortical thickness and gyrification compared to controls. Cortical thickness was not associated with immune health parameters, while gyrification was negatively associated with immune health measures. However, the linear rate of change of cortical thickness and gyrification from age 5 to 9 in the HIV+ children was not different from that of uninfected controls, nor was it different between controls and children on interrupted or continuous ART. Children with HIV-related encephalopathy showed a decrease in gyrification with age during this period, in contrast to controls who showed stable gyrification except in frontal regions where gyrification increased with age. Children with perinatal HIV infection display alterations in cortical development due to ART interruption and disease severity at age 5 years, despite starting ART early in life. Our results suggest that cortical gyrification is more sensitive than cortical thickness to effects of perinatal HIV infection. ART interruption and disease severity at ART initiation affect cortical morphometry development at age 5 years in a perinatally infected, early-treated pediatric cohort. However, on continued ART the cortical developmental trajectory is no different from that of uninfected controls. Any structural defects resulting from ART interruption appear to normalise by age 9, except in children with HIV-related encephalopathy, who show an altered trajectory of gyrification development.
  • Thumbnail Image
    Item
    Open Access
    Characterising the structural brain connectome in Alzheimer's disease and its relation to cognitive intra-individual variability.
    (2022) Lee, Shao-Hsuan Stephanie; Jankiewicz, Marcin; Robertson, Frances; Christ, Björn
    Research on Alzheimer's disease (AD) has shown white matter (WM) degeneration and structural connectome disruptions measured by diffusion tensor imaging (DTI) as well as increased reaction time intra-individual variability (RT IIV) on neurocognitive testing. However, the relationship between these changes measured on these different modalities remains unexplored. To explore possible relationships between these alterations, this study used tractography to identify WM changes, as indexed by fractional anisotropy (FA) and mean diffusivity (MD), followed by a graph theory-based analysis of the brain structural connectome, and investigated the relationship between these graph theory metrics and RT IIV in 16 AD patients and 20 healthy elderly controls. Within AD patients, we identified WM tracts with lower FA and higher MD mostly located in the cortical and subcortical temporal lobes, such as the hippocampus subregions, compared to healthy elderly controls. We also observed higher FA in the WM tracts within the thalamus as well as between the thalamus and brainstem in AD patients. In the structural brain connectome of these patients, there were regions with altered nodal strength, transitivity, and local efficiency relative to the controls' connectome. Conversely to many studies, we found increases in nodal efficiency across multiple regions and higher global efficiency in AD patients compared to healthy elderly controls. Finally, higher global efficiency was correlated with increased RT IIV on the CRT task in AD patients. In AD, a positive relationship between transitivity in the left cingulate cortex and RT IIV as well as between nodal efficiency in the left cortical temporal lobe and mean RT on the CRT task were observed. On the other hand, lower transitivity in the right thalamus and increased RT IIV, as well as lower transitivity and longer mean RT were found on both the SRT and CRT tasks in AD patients. Our results may show evidence of disruptions and compensatory mechanisms in WM tracts and network metrics in AD. Together, these results revealed WM changes, topological alterations of the brain structural connectome in AD and that these findings can be used in combination with IIV to predict cognitive decline or progression of AD.
  • Thumbnail Image
    Item
    Open Access
    Comparison of Magnetic Resonance Spectroscopy (MRS) data in children with and without HIV at 11-12 years
    (2020) Graham, Amy; Robertson, Frances; Meintjes, Ernesta
    Although HIV and antiretroviral drugs have been shown to cause damage in the brain, the long-term impacts of perinatal infection, early treatment and exposure in children at 11 years, remain unclear. The effects of HIV and antiretroviral therapy (ART), whilst indistinguishable, can be investigated at a chemical level through proton magnetic resonance spectroscopy (1H-MRS). Previous studies in children have largely focused on individual metabolite changes. However, several adult studies have now advanced beyond this to address patterns of metabolic activity that are altered with HIV infection. Using a 3T Skyra scanner, 136 children (76 HIV+, 30 HEU, 30 HU; 71 males) between the ages of 11.0- 12.5 years, and from a similar socioeconomic background, were scanned. In this study metabolite concentrations were quantified within the basal ganglia (BG), midfrontal gray matter (MFGM) and peritrigonal white matter (PWM). We utilised linear regression to investigate individual metabolite differences, comparing HIV-infected (HIV+) children from the Children with HIV Early Antiretroviral Therapy (CHER) trial, and HIV-exposed-uninfected (HEU) children, to HIV-unexposed (HU) children. Pearson's correlation analysis, factor analysis and logistic regression were then used to study alterations in metabolic patterns between HIV+ and HIV-uninfected (HIV-) children. Analysis of the data was carried out in R. We found elevated total choline in the BG (p = 0.03) and MFGM (p < 0.001) of HIV+ children, as well as reduced PWM total NAA (p = 0.03) and total creatine (p = 0.01). Altered metabolite concentrations were further observed in HEU children. Additionally, we identified a cross-regional coupling of choline which distinguishes HIV+ from HIV- children (p < 0.001). These findings indicate that multiregional inflammation and PWM axonal damage are occurring in HIV+ children at 11 years. Ultimately, the consequences of perinatal HIV acquisition, in spite of early treatment, continue to be seen at 11 years, as do the impacts of exposure.
  • Thumbnail Image
    Item
    Open Access
    Double volumetric navigators for real-time simultaneous shim and motion measurement and correction in Glycogen Chemical Exchange Saturation Transfer (GlycoCEST) MRI
    (2018) Simegn, Gizeaddis Lamesgin; Robertson, Frances
    Glycogen is the primary glucose storage mechanism in in living systems and plays a central role in systemic glucose homeostasis. The study of muscle glycogen concentrations in vivo still largely relies on tissue sampling methods via needle biopsy. However, muscle biopsies are invasive and limit the frequency of measurements and the number of sites that can be assessed. Non-invasive methods for quantifying glycogen in vivo are therefore desirable in order to understand the pathophysiology of common diseases with dysregulated glycogen metabolism such as obesity, insulin resistance, and diabetes, as well as glycogen metabolism in sports physiology. Chemical Exchange Saturation Transfer (CEST) MRI has emerged as a non-invasive contrast enhancement technique that enables detection of molecules, like glycogen, whose concentrations are too low to impact the contrast of standard MR imaging. CEST imaging is performed by selectively saturating hydrogen nuclei of the metabolites that are in chemical exchange with those of water molecules and detecting a reduction in MRI signal in the water pool resulting from continuous chemical exchange. However, CEST signal can easily be compromised by artifacts. Since CEST is based on chemical shift, it is very sensitive to field inhomogeneity which may arise from poor initial shimming, subject respiration, heating of shim iron, mechanical vibrations or subject motion. This is a particular problem for molecules that resonate close to water, such as - OH protons in glycogen, where small variations in chemical shift cause misinterpretation of CEST data. The purpose of this thesis was to optimize the CEST MRI sequence for glycogen detection and implement a real-time simultaneous motion and shim correction and measurement method. First, analytical solution of the Bloch-McConnell equations was used to find optimal continuous wave RF pulse parameters for glycogen detection, and results were validated on a phantom with varying glycogen concentrations and in vivo on human calf muscle. Next, the CEST sequence was modified with double volumetric navigators (DvNavs) to measure pose changes and update field of view and zero- and first-order shim parameters. Finally, the impact of B0 field fluctuations on the scan-rescan reproducibility of CEST was evaluated in vivo in 9 volunteers across 10 different scans. Simulation results showed an optimal RF saturation power of 1.5µT and duration of 1s for glycoCEST. These parameters were validated experimentally in vivo and the ability to detect varying glycogen concentrations was demonstrated in a phantom. Phantom data showed that the DvNav-CEST sequence accurately estimates system frequency and linear shim gradient changes due to motion and corrects resulting image distortions. In addition, DvNav-CEST was shown to yield improved CEST quantification in vivo in the presence of motion and motion-induced field inhomogeneity. B0 field fluctuations were found to lower the reproducibility of CEST measures: the mean coefficient of variation (CoV) for repeated scans was 83.70 ± 70.79 % without shim correction. However, the DvNav-CEST sequence was able to measure and correct B0 variations, reducing the CoV to 2.6 ± 1.37 %. The study confirms the possibility of detecting glycogen using CEST MRI at 3 T and shows the potential of the real-time shim and motion navigated CEST sequence for producing repeatable results in vivo by reducing the effect of B0 field fluctuations.
  • No Thumbnail Available
    Item
    Open Access
    Effects of HIV on longitudinal development of the fronto-striatal resting state networks in children from 7 to 11 years
    (2023) Brown, Cayleigh; Robertson, Frances; Meintjes Ernesta
    Despite the early initiation of antiretroviral therapy (ART), human immunodeficiency virus (HIV) in children is still associated with neurodevelopmental delay, with findings of neurocognitive deficits in executive, motor and language function. The basal ganglia (BG) develop early in childhood and are known to be affected by HIV. Their main function is to control goal-directed behaviours. Although the BG are strongly interconnected with the cerebral cortex through fronto-striatal pathways, little is known about frontostriatal network maturation during childhood or how perinatal HIV-infection (PHIV) affects its development. Using seed-based correlation analysis (SCA) of resting state functional magnetic resonance imaging (rs-fMRI), we conducted a longitudinal analysis to examine the effects of age and HIV on functional connectivity (FC) between the BG and the frontal cortex. Rs-fMRI data were collected at ages 7, 9 and 11 from children with perinatal HIV (CPHIV) who were part of the Children with HIV Early AntiRetroviral (CHER) trial (n=79), along with age-matched uninfected children from similar neighbourhoods (n=80). For SCA, BG seeds and frontal cortex regions of interest (ROIs) were selected from known cortico-BGthalamic functional networks. The BG seeds comprised the bilateral putamen, caudate, globus pallidus (GP) and thalamus. Functional regions located in the dorsolateral prefrontal cortex (DLPFC), paracentral lobule and pars opercularis were chosen as frontal ROIs because of their association with executive, motor and language function, respectively. FC was obtained via correlations between rs-fMRI time-series from the BG seeds and frontal ROIs. Linear mixed effect models were used to examine individual and synergistic effects of age and HIV status on FC. We found age-related increases in FC between BG seeds and both DLPFC and pars opercularis regions, while age-related FC decreases were seen between the BG and paracentral lobule. HIV-related alterations in FC were seen between the paracentral lobule and left and right caudate (P = 0.043 and P = 0.02) and GP (P = 0.023 and P = 0.041), and between the right pars opercularis and left and right putamen (P = 0.015 and P = 0.01), left caudate (P = 0.02) and left and right GP (P = 0.001 and P = 0.048). Attenuated FC between the BG and paracentral lobule and right pars opercularis suggests that HIV alters primarily motor function and inhibitory control, a component of executive function, in childhood. In contrast, there was no effect of HIV on FC of the BG to the DLPFC, a connection which is also considered to be important for executive functioning. Interaction of HIV and age on FC between the BG and paracentral lobule shows that although FC typically decreases with age, it remains unchanged in CPHIV. Similarly, increased FC between the BG and right pars opercularis with age in the control group but relatively constant FC in CPHIV provides evidence that HIV may hinder typical age-related development of BG FC.
  • Thumbnail Image
    Item
    Open Access
    Issues in the processing and analysis of functional NIRS imaging and a contrast with fMRI findings in a study of sensorimotor deactivation and connectivity
    (2012) Robertson, Frances; Meintjies, Ernesta; Douglas, Tania S
    The first part of this thesis examines issues in the processing and analysis of continuous wave functional linear infrared spectroscopy (fNIRS) of the brain usung the DYNOT system. In the second part, the same sensorimotor experiment is carried out using functional magnetic resonance imaging (fMRI) and near infrared spectroscopy in eleven of the same subjects, to establish whether similar results can be obtained at the group level with each modality. Various techniques for motion artefact removal in fNIRS are compared. Imaging channels with negligible distance between source and detector are used to detect subject motion, and in data sets containing deliberate motion artefacts, independent component analysis and multiple-channel regression are found to improve the signal-to-noise ratio.
  • Thumbnail Image
    Item
    Open Access
    Longitudinal analysis of Brain Metabolite levels for HIV infected Children from ages five to eleven
    (2020) Van Biljon, Noëlle; Little, Francesca; Meintjes, Ernesta; Holmes, Martha; Robertson, Frances
    HIV infected (HIV+) children initiate antiretroviral therapy (ART) early in life and remain on it lifelong. However, the long-term impact of ART and HIV on the maturing brain is not well documented and longitudinal neuroimaging studies are rare, especially in developing countries most heavily impacted by HIV/AIDS where access to imaging resources are limited. We have examined HIV related changes in metabolite level trajectories from 5-11 years in three brain regions using Magnetic Resonance Spectroscopy (MRS). We used univariate linear mixed effect models to identify independent profiles of the metabolites measured in each region of the brain. To explore the metabolite trends in a multivariate setting we generated multilevel mixed effects models, and correlated response models. There was an element of confounding introduced through the change of MRI scanner during the follow-up period and we compare different methods to resolve this issue. Consequently, we did observe differences in metabolite profiles from HIV+ children compared to HIV uninfected (HIV-) controls. This suggests that while these children are on ART treatment, there is still some underlying effect on their neurochemistry which sets their development apart from the normal healthy profiles we expect.
  • Thumbnail Image
    Item
    Open Access
    Multimodal neuroimaging signatures of early cART-treated paediatric HIV - Distinguishing perinatally HIV-infected 7-year-old children from uninfected controls
    (2020) Khobo, Isaac Lebogang; Robertson, Frances; Jankiewicz, Marcin; Holmes, Martha
    Introduction: HIV-related brain alterations can be identified using neuroimaging modalities such as proton magnetic resonance spectroscopy (1H-MRS), structural magnetic resonance imaging (sMRI), diffusion tensor imaging (DTI), and functional MRI (fMRI). However, few studies have combined multiple MRI measures/features to identify a multivariate neuroimaging signature that typifies HIV infection. Elastic net (EN) regularisation uses penalised regression to perform variable selection, shrinking the weighting of unimportant variables to zero. We chose to use the embedded feature selection of EN logistic regression to identify a set of neuroimaging features characteristic of paediatric HIV infection. We aimed to determine 1) the most useful features across MRI modalities to separate HIV+ children from HIV- controls and 2) whether better classification performance is obtained by combining multimodal MRI features rather than using features from a single modality. Methods: The study sample comprised 72 HIV+ 7-year-old children from the Children with HIV Early Antiretroviral Therapy (CHER) trial in Cape Town, who initiated combination antiretroviral therapy (cART) in infancy and had their viral loads suppressed from a young age, and 55 HIV- control children. Neuroimaging features were extracted to generate 7 MRI-derived sets. For sMRI, 42 regional brain volumes (1st set), mean cortical thickness and gyrification in 68 brain regions (2nd and 3rd set) were used. For DTI data: radial (RD), axial (AD), mean (MD) diffusivities, and fractional anisotropy (FA) in each of 20 atlas regions were extracted for a total of 80 DTI features (4th set). For 1H-MRS, concentrations of 14 metabolites and their ratios to creatine in the basal ganglia, peritrigonal white matter, and midfrontal gray matter voxels (5th, 6th and 7th set) were considered. A logistic EN regression model with repeated 10-fold cross validation (CV) was implemented in R, initially on each feature set separately. Sex, age and total intracranial volume (TIV) were included as confounders with no shrinkage penalty. For each model, the classification performance for HIV+ vs HIV- was assessed by computing accuracy, specificity, sensitivity, and mean area under the receiver operator characteristic curve (AUC) across 10 CV folds and 100 iterations. To combine feature sets, the best performing set was concatenated with each of the other sets and further EN regressions were run. The combination giving the largest AUC was combined with each of the remaining sets until there was no further increase in AUC. Two concatenation techniques were explored: nested and non-nested modelling. All models were assessed for their goodness of fit using χ 2 likelihood ratio tests for non-nested models and Akaike information criterion (AIC) for nested models. To identify features most useful in distinguishing HIV infection, the EN model was retrained on all the data, to find features with non-zero weights. Finally, multivariate imputation using chained equations (MICE) was explored to investigate the effect of increased sample size on classification and feature selection. Results: The best performing modality in the single modality analysis was sMRI volumes
  • Thumbnail Image
    Item
    Open Access
    Segmentation in image sequences : tracking human figures in motion
    (2001) Robertson, Frances; De Jager, Gerhard
    This thesis addressed the issue of segmentation and tracking of human figures in video sequences. Various issues affecting segmentation are investigated, including selection of a colour space representation. Different methods for segmentation are implemented and their respsective merits and disadvantages discovered. These are largely various forms of colour and motion segmentation. The different forms are then combined in order to produce a segmentation that is more effective than any of the component techniques on its own. The tracking problem is addressed separately through the use of a Smart Room, where the segmentation can be easily performed. The tracking involves a trajectory for a given person over time in the image frame co-ordinate system.