The BrainSuite Statistics Toolbox in R (bssr)

Presented During:


Poster No:

1191 

Submission Type:

Abstract Submission 

Authors:

Shantanu Joshi1, Yeun Kim1, Kayla Schroeder1, Anand Joshi2, Richard Leahy2, David Shattuck3

Institutions:

1UCLA, Los Angeles, CA, 2University of Southern California, Los Angeles, CA, 3University of California, Los Angeles, Los Angeles, CA

First Author:

Shantanu Joshi  
UCLA
Los Angeles, CA

Co-Author(s):

Yeun Kim  
UCLA
Los Angeles, CA
Kayla Schroeder  
UCLA
Los Angeles, CA
Anand Joshi  
University of Southern California
Los Angeles, CA
Richard Leahy  
University of Southern California
Los Angeles, CA
David Shattuck, PhD  
University of California, Los Angeles
Los Angeles, CA

Introduction:

The BrainSuite Statistics toolbox in R (bssr) is a software package developed in R that performs statistical analysis of population-level neuroimaging data processed using BrainSuite [1]. Specifically, it provides statistical tools for conducting cortical thickness analysis, tensor based morphometry, and analysis of diffusion measures.

Methods:

A subject-level BrainSuite workflow prior to conducting statistical analysis involves T1-weighted MRI image processing and registration steps include cortical surface extraction [1] and alignment to a reference atlas using SVReg [2]. SVReg performs surface-constrained volume registration of triangular meshes and image intensities. Bssr is then used to perform population level statistical analysis of various neuroimaging measures (Figure 1).

Bssr supports the following analysis methods: i) tensor based morphometry (TBM) analysis of voxel-wise magnitudes of the 3D deformation fields of MRI images registered to the atlas; ii) cortical surface analysis of the vertex-wise thickness in the atlas space, iii) diffusion parameter maps analysis (e.g., fractional anisotropy, mean diffusivity, radial diffusivity). The statistical analysis is performed in a common coordinate space of an atlas by resampling the data from subject coordinates to a common atlas space using SVReg [2,3]; and iv) region of interest (ROI)-based analysis of average gray matter thickness, surface area, and gray matter volume within cortical ROIs. It also offers tools for correcting for multiple comparisons using false discovery rate (FDR) or permutation testing methods.

Bssr is cross-platform and is available on macOS, Windows,and Linux based systems (all platforms with R support) is distributed under an open source license (GPLv2). Bssr supports functionality for automated report generation to visualize statistical results using R-shiny and R markdown. The volumetric analysis report contains the cluster table, visualizations of clusters on image slices, and shows both the unadjusted and the adjusted versions of p-values and t statistics respectively. The ROI analysis report shows the demographic spreadsheet, automatic bar plots for ANOVA and regressions, and scatter plot for correlation analyses. Bssr also exports an R markdown report that contains reproducible R commands in both the Rmd file, and in the html document [4]. This enables complete reproducibility of statistical results and only requires packaging the R markdown file along with the data.
Supporting Image: Fig1_bssr_workflow.png
 

Results:

We demonstrate bssr on MRI analyses on a subsample of the Rockland dataset (N=25 healthy controls, ages 18-75 years, 7 M/18F) [5]. Results are shown in Figure 2 in the form of screenshots from the automated web-based statistical reporting tool. Figure 2A shows correlation of age with the magnitude of volumetric shrinking and expansion in the brain (TBM). Figure 2B shows an ANOVA analysis of the effects of age on cortical thickness, whereas Figure 2C shows the ANOVA table from an ROI analysis of the age effect on the right pars opercularis. All results were controlled for sex and were corrected for multiple comparisons using FDR; alpha=0.05. Even in this small sample, we observe significant correlations of age with the volumetric expansion of the ventricles and shrinking in the prefrontal cortex for the TBM analysis (Fig 2B), and cortical atrophy in the motor cortex, parietal lobes, and in the superior temporal gyrus (Fig 2B). The ROI analysis shows a significant (p=0.0495, uncorrected) negative correlation with the right pars opercularis in the inferior frontal gyrus (Fig 2C).
Supporting Image: Fig2_results.png
 

Conclusions:

The R development environment for bssr not only enables reproducible statistical models for statistical analysis, but also offers the end user full interoperability with existing R software packages including those provided by Neuroconductor [6]. In the future, we plan to expand bssr to include connectivity analyses of neuroimaging data in BrainSuite. Bssr is available to download from http://brainsuite.org/bssr.

Modeling and Analysis Methods:

Methods Development
Univariate Modeling 1

Neuroanatomy, Physiology, Metabolism and Neurotransmission:

Neuroanatomy Other

Neuroinformatics and Data Sharing:

Workflows
Informatics Other 2

Keywords:

Computing
Data analysis
Design and Analysis
Modeling
Morphometrics
Multivariate
Statistical Methods

1|2Indicates the priority used for review

My abstract is being submitted as a Software Demonstration.

Yes

Please indicate below if your study was a "resting state" or "task-activation” study.

Other

Healthy subjects only or patients (note that patient studies may also involve healthy subjects):

Healthy subjects

Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.

Yes

Was any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.

Not applicable

Was any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.

Not applicable

Please indicate which methods were used in your research:

Structural MRI
Diffusion MRI
Other, Please specify  -   bssr
Computational modeling

For human MRI, what field strength scanner do you use?

3.0T

Which processing packages did you use for your study?

Other, Please list  -   BrainSuite

Provide references using author date format

1. Shattuck DW et al. (2002) BrainSuite: An Automated Cortical Surface Identification Tool Medical Image Analysis, 8(2):129-142.

2. Joshi AA et al. (2007) Surface-Constrained Volumetric Brain Registration Using Harmonic Mappings IEEE Trans. on Medical Imaging 26(12):1657-1669.

3. Bhushan C et al. (2015) Co-registration and distortion correction of diffusion and anatomical images based on inverse contrast normalization. Neuroimage (7):115:269-80.

4. Xie Y (2017) Dynamic Documents with R and knitr. Chapman and Hall/CRC.

5. Nooner KB et al. (2012) The NKI-Rockland sample: a model for accelerating the pace of discovery science in psychiatry. Frontiers in neuroscience 6:152.

6. Muschelli J et al. (2018) Neuroconductor: an R platform for medical imaging analysis. Biostatistics 20.2: 218-239.