OpenUCT :: Browsing by Author "Meintjies, Ernesta"

Browsing by Author "Meintjies, Ernesta"

Now showing 1 - 9 of 9

Open Access
The effect of early adversity on adult spatial cognition : a functional magnetic resonance imaging study
(2009) Dey, Marc; Thomas, Kevin; Meintjies, Ernesta
Exposure to traumatic childhood events can lead to a range of behavioural, psychological, and physiological consequences. Previous studies have shown that neurobiological changes in reaction to severe stress may cause lasting damage to particular neural regions, including the hippocampus and the prefrontal cortex. It has been suggested that such damage to these regions results in difficulties in associated cognitive functioning, including problems with verbal declarative memory and cognitive control. Little focus has been placed on visualspatial cognition in traumatised individuals, however. The aim of this project, which comprised two studies, was to investigate visual memory and spatial cognition in adult survivors of childhood trauma. Study 1 compared the performance of 23 individuals who had experienced childhood abuse (the Trauma group) to 38 matched controls with no such experience (the Control group) on the four visual-spatial memory tasks of the Cambridge Neuropsychological Test Automated Battery (CANTAB). Results suggested that participants in the Trauma group showed poorer performance on two of the more complex tasks, which tapped both hippocampal and prefrontal cortex functioning, compared to the controls. One interpretation of this finding is that these between-group differences reflect the dysfunction of a network involved in visual-spatial memory in individuals who have experienced childhood trauma. Study 2 used functional magnetic resonance imaging (fMRI) to investigate whether any marked differences in neural activation would be evident between individuals with a history of childhood trauma (n = 7) and matched controls with no such history (n = 14) during spatial navigation tasks. Functional images were gathered while participants completed two spatial navigation tasks: the Computer-Generated Arena (CG Arena), a smallscale spatial navigation task, and the Virtual City, a large-scale spatial navigation task based on an environment created by Maguire et al. (1998). Although no significant behavioural differences were evident during the completion of these tasks, the fMRI data did show marked differences in activation. These results of the CG Arena, in particular, showed lower activation in PFC areas, including the anterior cingulate cortex, during wayfinding tasks. Taken together, the results of these two studies suggest that (a) subtly impaired neural functioning is evident in individuals with a history of childhood trauma, and (b) this impairment may lead to difficulties in successfully completing complex visual-spatial memory and spatial navigation tasks.
Open Access
Implementation of Hadamard Encoding and Reconstruction of MEGA-Edited Spectroscopy (HERMES) for quantification of ɣ-aminobutyric acid (GABA) and glutathione (GSH)
(2020) Rimbault, Daniel; Meintjies, Ernesta; Alhamud, Ali; Edden, Richard
The present study aimed to accelerate and improve accuracy of ɣ-aminobutyric acid (GABA) and glutathione (GSH) quantification. These metabolites, present at low concentrations in the brain, are challenging to detect using MR spectroscopy due to the fact that their resonance frequencies overlap with those of other more abundant metabolites. The advanced spectral editing techniques involving J-difference editing that are required to resolve the overlapping signals of these low concentration metabolites are not routinely available on clinical MRI scanners. In this work we implemented on a 3T Siemens Skyra MRI a novel MRS technique called Hadamard Encoding and Reconstruction of MEGA-Edited Spectroscopy (HERMES) to simultaneously detect GABA and GSH, developed a novel postprocessing technique that simultaneously models the sum and various difference spectra, and evaluated the performance of the sequence and processing method both in phantoms and in vivo. HERMES was implemented by modifying the Siemens GABA-edited MEGA-PRESS WIP sequence to include two additional sub-experiments – one editing GSH with a single lobe pulse and one editing both GABA and GSH using a dual lobe pulse, and replacing the Siemens pulses with ‘universal' pulses similar to those used in a previous implementation of HERMES on a Philips platform. Performance was assessed in a phantom and 22 healthy adults, 15 of whom provided usable data (7 male, mean age 25.6 ± 2.7 yr). Three of the subjects were scanned 3 times to assess reproducibility. Data were processed and compared using a set of custom scripts in MATLAB. Following frequency and phase correction of individual averages with GANNET, we applied our custom simultaneous linear combination model that iteratively fit the concatenated sum and difference spectra using a least squares routine. SPM was used for tissue segmentation of structural images and FID-A to simulate high-resolution basis sets. The simultaneous modelling technique provided absolute quantification results for 15 metabolite moieties using internal unsuppressed water as a reference. The performance of the simultaneous fitting approach was compared to multiple independent fittings for HERCULES (Hadamard Editing Resolves Chemicals Using Linear-combination Estimation of Spectra) data that had been previously acquired on a 3T Philips Achieva MRI. Although the HERMES sequence implemented on the Siemens platform as part of this project was able to successfully edit both GABA and GSH, and generate line shapes consistent with the work by Saleh et al. (2016), quantification accuracy was disappointing. In the phantom data, GSH and GABA concentrations were both roughly 50% of known levels. Since the actual concentrations in vivo were not known, we were not able to establish accuracy, but quantification agreement between the MEGA-PRESS and HERMES sequences was poor for most metabolites. Specifically, GABA levels were two to three times higher than expected values using both HERMES and GABA-edited MEGA-PRESS. Despite poor absolute agreement, concentrations from HERMES and MEGA-PRESS data were moderately correlated, and HERMES data showed lower coefficients of variation across subjects, suggesting that it may be more reliable. HERMES was also more reproducible across scanning sessions and participants for more metabolites than GABA- or GSH-edited MEGA-PRESS. Our findings also showed that simultaneous fitting using the sum and difference spectra produces lower coefficients of variation for most metabolites than fittings to sum and difference spectra separately.
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.
Open Access
Magnetic resonance spectroscopy quality assessment at CUBIC and application to the study of the cerebellar deep nuclei in children with fetal alcohol spectrum disorder
(2010) Du Plessis, Lindie; Meintjies, Ernesta
In vivo magnetic resonance spectroscopy (MRS) is an imaging technique that allows the chemical study of human tissue non-invasively. The method holds great promise as a diagnostic tool once its reliability has been established. Inter-scanner variability has, however, hampered this from happening as results cannot easily be compared if acquired on different scanners. In this study a phantom was constructed to determine the localisation efficiency of the 3 T Siemens Allegra MRI scanner located at the Cape Universities Brain Imaging Centre (CUBIC). Sufficient localisation is the key to acquiring useful spectroscopic data as only the signal from a small volume of interest (VOI) is typically acquired. The phantom consisted of a Perspex cube located inside a larger Perspex sphere. Solutions of the cerebral metabolites N-acetyl aspartate (NAA) and choline (Cho) were placed in the inner cube and outer sphere respectively. The phantom was scanned at a range of voxel sizes and echo times in order to determine parameters that typically indicate the performance of the scanner in question. The resultant full width at half maximum (FWHM) and signal to noise ratio (SNR) values indicated that optimal results were obtained for a voxel with dimensions 20 x 20 x 20 mm3. The selection efficiency could not be measured due to limitations in the scanner, but two other performance parameters ' extra volume suppression (EVS) and contamination ' could be determined. The EVS showed that the scanner was able to eliminate the entire background signal from the out-of-voxel region when voxel sizes with dimensions (20 mm)3 and (30 mm)3 were used. This performance decreased to 96.2% for a voxel size of (50 mm)3. The contamination indicated that the unwanted signal, weighted by the respective proton densities of the chemicals, ranged from 12% in the (20 mm)3 voxel to 24% in the (50 mm)3 voxel. These ranges are well within acceptable limits for proton MRS. Analysis of the water suppression achieved in the scanner showed an efficiency of 98.84%, which is acceptable for proton spectroscopy. It was also found that manual iv shimming of the scanner improved the spectra obtained, as compared to the automated shimming performed by the scanner. The second objective of the study was to quantify absolute metabolite concentrations in the familiar SI units of mM as results were previously mostly expressed as metabolite ratios. The LCModel software was used to assess two methods of determining absolute metabolite concentrations and the procedure using water scaling consistently showed superior performance to a method using a calibration factor. The method employing water scaling was then applied to a study of fetal alcohol spectrum disorder (FASD) where the deep cerebellar nuclei of children with FASD and a control group were scanned. The cerebellar nuclei were of interest as children with FASD show a remarkably consistent deficit in eye blink conditioning (EBC). The cerebellar deep nuclei is known to play a critical role in the EBC response. The results show significant decreases in the myo-inositol (mI) and total choline (tCho) concentrations of children with FASD in the deep cerebellar nuclei compared to control children. The FAS/PFAS subjects have a mean mI concentration of 4.6 mM as compared to a mean of 5.3 mM in the controls. A Pearson correlation showed that there was a significant relationship between decreasing mI concentrations with increasing prenatal alcohol exposure. The mean tCho concentrations are 1.3 mM for FAS/PFAS and 1.5 mM for the controls. There was no significant differences between the heavily exposed group and either the FAS/PFAS or the control subjects for either metabolite. The decreased mI and tCho concentrations may indicate deficient calcium signalling or decreased cell membrane integrity ' both of which can explain the compromised cerebellar learning in FASD subjects.
Open Access
Methods and adaptations required to perform small-animal MRI scanning using a large bore clinical MRI
(2012) Saleh, Muhammad G; Meintjies, Ernesta; Davies, Neil; Franz, Thomas
Small-animal imaging has been widely implemented to study succession of disease, therapeutic treatments and the effects of environmental insults. The gold standard noninvasive technique for following progression of heart failure in small-animal models is magnetic resonance imaging (MRI). The aim of this project was to adapt a clinical MRI system to perform small-animal cardiac MRI. The first part of the thesis describes the adaptations required, which included design and construction of a small-animal radiofrequency (RF) coil, physical support (cradle), a core body temperature regulation system, and optimization of pulse sequences. The system was validated using a phantom and in-vivo in 5 healthy rats. The signal-to-noise ratio (SNR) in the phantom was 91% higher using the small-animal coil compared to the standard head coil. SNRs of 7 ± 2 and 18.9 ± 0.6 were achieved in myocardium and blood, respectively, in healthy rats and MR left ventricular mass (LVM) was highly correlated with (r=0.87) with post-mortem mass. In the second part of the study, left ventricular remodeling (LVR) was investigated in a nonreperfused model of myocardial infarction (MI) in 5 sham and 7 infarcted rats. Rats were scanned at 2 and 4 weeks post surgery to allow for global and regional functional and structural analyses of the heart. Images were of sufficient quality to enable semi-automatic segmentation using Segment. Significant increase in end-systolic volume (ESV) was observed in MI rats at 2 weeks post surgery. At 4 weeks post surgery, end-diastolic volume (EDV) and ESV of MI rats were significantly higher than in sham rats. Ejection fraction (EF) of MI rats dropped significantly at 2 weeks and a further significant drop was observed at 4 weeks indicating contractile dysfunction. Wall thickness (WTh) analyses in MI rats at 4 weeks revealed significant reduction in end-diastolic (ED) wall thickness in the anterior region due to necrosis of myocytes. In the posterior region, WTh was significantly higher due to LV hypertrophy. At end-systole (ES), the MI rats revealed significant decrease in WTh in the anterior and lateral regions. MI rats suffered reduction in systolic wall thickening in all regions of the heart, indicating global contractile dysfunction.
Open Access
Noise reduction during diffusion tensor imaging of infants
(2019) Jordaan, Johannes Petrus; Meintjies, Ernesta; van der Kouwe, Andr´e
Acoustic noise produced during echo planar imaging (EPI) has been known to reach excessive levels. In addition to causing general patient discomfort and anxiety, this level of noise makes the scanning of young children and infants particularly difficult. Infants are typically scanned while sleeping to minimise motion as they cannot ethically be sedated for research purposes. The extreme noise during MRI acquisitions often cause them to wake before the end of the scanning session. This problem is exacerbated by particularly noisy acquisitions, such as the single shot echo planar diffusion tensor imaging sequence. The main aim of this project was to reduce the noise of this particular acquisition specifically for the scanning of newborn infants. Acoustic noise during MRI acquisitions mainly originates from mechanical vibrations in the gradient coil assemblies due to interactions between the rapidly changing currents applied to the coils and the main static field. A transfer function relating the output acoustic noise spectrum to the gradient excitation input spectrum was developed and used to identify resonant peaks which would amplify coinciding gradient waveform harmonics. In addition to resonant peaks, the transfer function showed significant amplification of frequencies above 1 kHz. In this work, noise reduction was achieved by implementing digital low-pass filters to reduce high-frequency harmonics of the standard trapezoidal gradient waveforms, focusing on the EPI readout portion of a diffusion tensor imaging (DTI) sequence. For comparison purposes, an EPI readout using sinusoidal frequency encoding waveforms and a constant phase encoding blip was also implemented. In addition to reducing produced noise, a passive noise reduction enclosure was built from open cell polyurethane foam mounted in a PVC frame to surround the sleeping infant and act as an acoustic insulation box. Lastly, the effectiveness of introducing pink noise from an external source to mask the abrupt changes in scanner noise, was also investigated. The altered k-space trajectories due to the modifications made to the EPI readout gradient waveforms were corrected through a custom one-dimensional regridding procedure applied along the frequency encoding axis in k-space. Noise reduction was measured with an Optimic 1155 optical microphone from Optoacoustics, attached on top of a cylindrical water phantom inside a 16 channel infant head coil in the isocenter along the Z-direction, facing the bore in the rightleft direction (similar to the orientation that the ears of a sleeping infant would be). Signal-to-noise ratio (SNR) and fractional anisotropy (FA) within the corpus callosum (CC) were compared for images acquired using the standard and modified (filtered and sinusoidal readouts) DTI sequences, the latter each for regridding kernel window sizes of 2 and 4, respectively. The acoustic noise spectra of the filtered and sinusoidal EPI sequences demonstrated a significant reduction in EPI harmonics compared to the standard sequence, but very little difference between each other. Without the foam enclosure, the filtered acquisition with filtered crushers reduced peak sound pressure levels (SPL) by 3.4 and 4 dBA for strong and no fat suppression, respectively, and A-weighted equivalent continuous sound levels (LA,eq) by 2.5 and 2.8 dBA, respectively. Adding the foam enclosure increased peak SPL reduction to 4.8 dBA with fat suppression and 7 dBA without. The sinusoidal sequence performing similarly or marginally (no more than 0.5%) worse than the filtered on all outcomes. SNR measurements in the CC were higher for all volumes of the filtered acquisition compared to the standard, while those of the sinusoidal were similar or slightly lower compared to the standard acquisition. FA values in the CC of the sinusoidal and filtered acquisitions did not differ from those of the standard acquisition (pairwise student’s t-test, all p’s >0.2). For the 16 channel head coil, image reconstruction time increased by only 45 seconds for a regridding kernel width W = 2. Filtering gradient waveforms is an effective technique to reduce acoustic noise during DTI without increasing acquisition time, reducing image quality, or altering FA measures. The proposed method has the potential to be generalized to most gradient waveforms across a variety of sequences. With the addition of the passive noise reduction enclosure, the combined noise reduction could greatly reduce infant anxiety and startling, leading to an increase in the number of infants in whom the acquisition protocol is completed. Acoustic noise produced during echo planar imaging (EPI) has been known to reach excessive levels. In addition to causing general patient discomfort and anxiety, this level of noise makes the scanning of young children and infants particularly difficult. Infants are typically scanned while sleeping to minimise motion as they cannot ethically be sedated for research purposes. The extreme noise during MRI acquisitions often cause them to wake before the end of the scanning session. This problem is exacerbated by particularly noisy acquisitions, such as the single shot echo planar diffusion tensor imaging sequence. The main aim of this project was to reduce the noise of this particular acquisition specifically for the scanning of newborn infants. Acoustic noise during MRI acquisitions mainly originates from mechanical vibrations in the gradient coil assemblies due to interactions between the rapidly changing currents applied to the coils and the main static field. A transfer function relating the output acoustic noise spectrum to the gradient excitation input spectrum was developed and used to identify resonant peaks which would amplify coinciding gradient waveform harmonics. In addition to resonant peaks, the transfer function showed significant amplification of frequencies above 1 kHz. In this work, noise reduction was achieved by implementing digital low-pass filters to reduce high-frequency harmonics of the standard trapezoidal gradient waveforms, focusing on the EPI readout portion of a diffusion tensor imaging (DTI) sequence. For comparison purposes, an EPI readout using sinusoidal frequency encoding waveforms and a constant phase encoding blip was also implemented. In addition to reducing produced noise, a passive noise reduction enclosure was built from open cell polyurethane foam mounted in a PVC frame to surround the sleeping infant and act as an acoustic insulation box. Lastly, the effectiveness of introducing pink noise from an external source to mask the abrupt changes in scanner noise, was also investigated. The altered k-space trajectories due to the modifications made to the EPI readout gradient waveforms were corrected through a custom one-dimensional regridding procedure applied along the frequency encoding axis in k-space. Noise reduction was measured with an Optimic 1155 optical microphone from Optoacoustics, attached on top of a cylindrical water phantom inside a 16 channel iv infant head coil in the isocenter along the Z-direction, facing the bore in the rightleft direction (similar to the orientation that the ears of a sleeping infant would be). Signal-to-noise ratio (SNR) and fractional anisotropy (FA) within the corpus callosum (CC) were compared for images acquired using the standard and modified (filtered and sinusoidal readouts) DTI sequences, the latter each for regridding kernel window sizes of 2 and 4, respectively. The acoustic noise spectra of the filtered and sinusoidal EPI sequences demonstrated a significant reduction in EPI harmonics compared to the standard sequence, but very little difference between each other. Without the foam enclosure, the filtered acquisition with filtered crushers reduced peak sound pressure levels (SPL) by 3.4 and 4 dBA for strong and no fat suppression, respectively, and A-weighted equivalent continuous sound levels (LA,eq) by 2.5 and 2.8 dBA, respectively. Adding the foam enclosure increased peak SPL reduction to 4.8 dBA with fat suppression and 7 dBA without. The sinusoidal sequence performing similarly or marginally (no more than 0.5%) worse than the filtered on all outcomes. SNR measurements in the CC were higher for all volumes of the filtered acquisition compared to the standard, while those of the sinusoidal were similar or slightly lower compared to the standard acquisition. FA values in the CC of the sinusoidal and filtered acquisitions did not differ from those of the standard acquisition (pairwise student’s t-test, all p’s >0.2). For the 16 channel head coil, image reconstruction time increased by only 45 seconds for a regridding kernel width W = 2. Filtering gradient waveforms is an effective technique to reduce acoustic noise during DTI without increasing acquisition time, reducing image quality, or altering FA measures. The proposed method has the potential to be generalized to most gradient waveforms across a variety of sequences. With the addition of the passive noise reduction enclosure, the combined noise reduction could greatly reduce infant anxiety and startling, leading to an increase in the number of infants in whom the acquisition protocol is completed.
Open Access
Real-time motion and magnetic field correction for GABA editing using EPI volumetric navigated MEGA-SPECIAL sequence: Reproducibility and Gender effects
(2016) Saleh, Muhammad G; Meintjies, Ernesta; Alhamud, Alkathafi Ali
γ-aminobutyric acid (GABA) is the primary inhibitory neurotransmitter and is of great interest to the magnetic resonance spectroscopy (MRS) community due to its role in several neurological diseases and disorders. Since GABA acquisition without macromolecule contamination requires long scan times and strongly depends on magnetic field (B0) stability, it is highly susceptible to motion and B0 inhomogeneity. In this work, a pair of three-dimensional (3D) echo planar imaging (EPI) volumetric navigators (vNav) with different echo times, were inserted in MEGA-SPECIAL to perform prospective correction for changes in the subject's head position and orientation, as well as changes in B0. The navigators do not increase acquisition time and have negligible effect on the GABA signal. The motion estimates are obtained by registering the first of the pairs of successive vNav volume images to the first volume image. The 3D field maps are calculated through complex division of the pair of vNav contrasts and are used for estimating zero- and first-order shim changes in the volume of interest (VOI). The efficacy of the vNav MEGA-SPECIAL sequence was demonstrated in-vitro and in vivo. Without motion and shim correction, spectral distortions and increases in spectral fitting error, linewidth and GABA concentration relative to creatine were observed in the presence of motion. The navigated sequence yielded high spectral quality despite significant subject motion. Using the volumetric navigated MEGA-SPECIAL sequence, the reproducibility of GABA measurements over a 40 minute period was investigated in two regions, the anterior cingulate (ACC) and medial parietal (PAR) cortices, and compared for different analysis packages, namely LCModel, jMRUI and GANNET. LCModel analysis yielded the most reproducible results, followed by jMRUI and GANNET. GABA levels in ACC were unchanged over time, while GABA levels in PAR were significantly lower for the second measurement. In ACC, GABA levels did not differ between males and females. In contrast, males had higher GABA levels in PAR. This gender difference was, however, only present in the first acquisition. Only in males did GABA levels in PAR decrease over time. These results demonstrate that gender differences are regional, and that GABA levels may fluctuate differently in different regions and sexes.
Open Access
Real-time motion and main magnetic field correction in MR spectroscopy using an EPI volumetric navigator
(2011) Hess, Aaron T; Meintjies, Ernesta; Van der Kouwe, Andre
In population groups where subjects do not lie still during Magnetic Resonance Spectroscopy (MRS) scans, real-time volume of interest (VOI), frequency, and main magnetic field (B0) shim correction may be necessary. This work demonstrates firstly that head movement causes significant B0 disruption in both single voxel spectroscopy and spectroscopic imaging.
Open Access
Tissue Parameter Mapping in Children with Fetal Alcohol Spectrum Disorders
(2020) Fourie, Marilize; Jankiewicz, Marcin; Meintjies, Ernesta
Background: Fetal alcohol spectrum disorders (FASD), which are caused by prenatal alcohol exposure (PAE), affects people around the world. Certain communities in South Africa have among the highest reported incidences of fetal alcohol syndrome (FAS) in the world. Although PAE-related brain alterations have been widely documented, the mechanisms whereby alcohol affects the brain are not clearly understood. MRI relaxation parameters T1, T2, T2* and proton density (PD), are basic tissue properties that reflect the underlying biology. The present study aims to advance our understanding of how PAE alters the microstructural properties of tissue by examining PAE-related changes in these tissue parameters in adolescents with FASD. Methods: The final sample used in this study consisted of 53 children from a previously studied longitudinal cohort (Jacobson et al., 2008) and 12 additionally recruited subjects. Of the 65 participants, 18 were diagnosed with FAS or partial FAS (PFAS) and made up the FAS/PFAS group, 18 were diagnosed as heavily exposed non-syndromal (HE) and 29 were age matched controls. Subjects were scanned at the Cape Universities Body Imaging Centre (CUBIC) located at Groote Schuur Hospital on a 3T Siemens Skyra MRI. Structural images were obtained using the MEMPRAGE sequence. From these images T1, T2, T2* and PD parameter maps were constructed and segmented into 43 regions of interest (ROI) using Freesurfer, FSL and AFNI. Linear regression analyses were used to analyse group differences as well as correlations between parameter values and the amount of alcohol the mother consumed during pregnancy. Results: Significant T1 differences were found in the caudate, cerebellar cortex, hippocampus, accumbens, putamen, choroid plexus, ventral diencephalon (DC), right vessel and ventricles. Significant T2 differences were found in the caudate, brain stem, corpus callosum (CC), amygdala, cerebral cortex, choroid plexus, vessels and ventricles. Significant T2* differences were found in the cerebellar cortex, optic chiasm and ventricles. Significant PD differences were found in the hippocampus and left lateral ventricle. The exploratory nature of this study resulted in none of the results surviving FDR correction for multiple comparisons. Conclusions: Overall, our findings point to regional PAE-related increases in water content and cellular and molecular changes in underlying tissue of the anatomical structure. Exceptions were the right cerebral cortex, brain stem, hippocampus, amygdala and ventral diencephalon where our findings point to less free water and increased cell density, and cerebellar cortex where simultaneous reductions in T1 and T2* suggest the possibility of increased iron content. In highly myelinated white matter structures, such as the CC and optic chiasm, our results point to PAErelated demyelination, and possibly increased iron. These findings extend previous knowledge of effects of PAE and demonstrate that tissues are affected at a microstructural level.

Browsing by Author "Meintjies, Ernesta"

Results Per Page

Sort Options