Network-Based White Matter Microstructure Measures Using b-Tensor Encoding and Cognition in HIV
Poster No:
1612
Submission Type:
Abstract Submission
Authors:
Md Nasir Uddin1, Abrar Faiyaz1, Chase Figley2, Xing Qiu1, Miriam Weber1, Giovanni Schifitto1
Institutions:
1University of Rochester, Rochester, NY, 2University of Manitoba, Winnipeg, MB
First Author:
Co-Author(s):
Introduction:
Despite receiving combination antiretroviral therapy (cART), individuals infected with HIV (HIV+) persistently contend with chronic inflammation, consequently elevating the risk of brain injury[1,2]. Diffusion MRI is highly sensitive to microstructural damage, but standard read-outs such as DTI are hindered by factors like regional fiber orientation dispersion (e.g., crossing fibers), impeding the detection of specific regional or network-based pathology[3,4]. Recently introduced, b-tensor encoding serves as a method capable of mapping microscopic fractional anisotropy (µFA) independently of intricate fiber architecture, including crossing and bending fibers[5-7]. Moreover, the b-tensor protocol estimates tissue heterogeneity via the anisotropic kurtosis (MKa) and isotropic kurtosis (MKi). Prior studies in HIV have shown relationships between cognitive performance and resting state functional connectivity in several brain networks[8]. In this work, we examine the relationships between b-tensor metrics throughout the functionally defined brain networks with neurocognitive scores
Methods:
Fifty-five participants were evaluated (24 HIV+: mean age=55±15years, female=8; and 31 HIV- controls: mean age=55±10years, female=7). After obtaining written informed consent, all participants underwent scanning using a 3T Siemens MRI. The MRI protocol included: 3D MPRAGE and FLAIR (1mm isotropic resolution); diffusion weighted imaging (DWI) using 2D SE-EPI (1.5mm isotropic resolution, 128 gradients with b=1000 and 2000 and 7 b=0 s/mm2); 4) b-tensor encoding using a free waveform encoding sequence with linear and spherical tensor encodings for four b-values (i.e., 100,700,1400,2000 s/mm2), 80 non-colinear directions with resolution=2×2×4 mm3. DWI data were preprocessed using FSL[9]. The b-tensor data were preprocessed using Multidimensional diffusion MRI toolbox[12]. The UManitoba-JHU functionally defined human white matter (WM) atlases[13, 14] and b-tensor maps were nonlinearly transformed to DWI native space using ANTs[15]. We then extracted ROI means in WM pathways underlying 12 previously-reported intrinsic functional brain networks[16]. These include – dorsal and ventral Default Mode Network, left and right Executive Control Network, anterior and posterior Salience Network, Precuneus Network, Language Network, Basal Ganglia Network, Higher Visual Network, Visuospatial Network, and Sensorimotor Network. Two-sample t-tests were used for group comparisons. All cognitive testing was performed using a neuropsychological battery that included tests of attention and working memory, processing speed, executive function, fine motor skills, verbal and visual learning, verbal and visual memory and language. The effects of HIV status on cognitive impairment and their interactions with imaging metrics were investigated via two-way analysis of variance (ANOVA). In addition, Pearson partial correlations were used to examine linear associations between cognitive performance and imaging metrics, with age as a covariate. All statistical analyses were performed in Python, and p<0.05 was considered significant.
Results:
We observed lower µFA and MKa, along with higher MKi, in the HIV+ group compared to the HIV- group; however, these differences did not reach statistical significance. A two-way ANOVA examining HIV status and its interaction with imaging metrics revealed that the basal ganglia network (BGN) is the most affected among the 12 functionally defined networks investigated. µFA in BGN was strongly positively associated with cognitive function, especially in HIV- participants. This association is particularly pronounced and statistically significant in subdomains such as processing speed and executive function (Table 1, Fig. 1).
Conclusions:
b-tensor metrics can be used to evaluate abnormalities associated with HIV infection in white matter networks (particularly the basal ganglia network) and may overcome certain limitations of traditional DTI with regard to fiber architecture.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 2
Diffusion MRI Modeling and Analysis 1
Multivariate Approaches
Other Methods
Novel Imaging Acquisition Methods:
Imaging Methods Other
Keywords:
Basal Ganglia
Cognition
Data analysis
DISORDERS
Infections
MRI
STRUCTURAL MRI
White Matter
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
1|2Indicates the priority used for review
Provide references using author date format
2. Haddow, L.J., et al., AIDS research and human retroviruses, 2019.
3. Figley, C.R., et al., Front in Neuro, 2022. 15: p. 799576.
4. Jeurissen, B., et al., HBM, 2013. 34(11): p. 2747-2766.
5. Szczepankiewicz, F., et al. PloS one, 2019. 14(3): p. e0214238.
6. Nilsson, M., et al. MRM 83.2 (2020): 608-620.
7. Szczepankiewicz et al., Data in brief, 2019. 25:104208.
8. Zhuang, Y., et al., Journal of neurovirology, 2017. 23(5): p. 704-712.
9. Zhang, H., et al., Neuroimage, 2012. 61(4): p. 1000-16.
10. Klein, S., et al., IEEE transactions on medical imaging, 2009. 29(1): p. 196-205.
11. Nilsson, M., et al., PloS one, 2015. 10(11): p. e0141825.
12. Nilsson, M., et al., in Proc Intl Soc Mag Reson Med. 2018.
13. Figley, T.D., et al., Frontiers in human neuroscience, 2017. 11: p. 306.
14. Figley, T.D., et al., Frontiers in human neuroscience, 2015. 9: p. 585.
15. Avants, B.B., et al., Neuroimage, 2011. 54(3): p. 2033-2044.
16. Shirer, W.R., et al., Cerebral cortex, 2012. 22(1): p. 158-165.