About Us

John Muschelli

Assistant Scientist, Department of Biostatistics

PhD in Biostatistics

  • worked on fMRI, sMRI, and CT for ≈ 7 years

Email: [email protected]

Kristin Linn

Assistant Professor, Department of Biostatistics and Epidemiology Perelman School of Medicine, University of Pennsylvania

PhD in Statistics

  • worked on sMRI for 3 years

Email: [email protected]

Overall Pipeline

Imaging Data Used: Multiple Sclerosis

  • Multiple sclerosis (MS) is a chronic disease of the central nervous system (brain, spinal chord, optic nerves)
  • MS lesions in the brain are areas of active inflammation, demylenation, or permanent tissue damage.
    • lesions are primarily in the white matter
  • The full data is available at https://smart-stats-tools.org.

Imaging Data Used: Multiple Sclerosis

  • MRI is well-suited for assessing lesion burden (volume and patterns) because lesions appear as hyperintensities on FLAIR and T2-w images and as hypointensities on T1-w images.
  • Using 5 training and 3 test subjects data from the an open MS data set (http://lit.fe.uni-lj.si/tools.php?lang=eng) (Lesjak et al. 2017)

Formats of Images

There are multiple imaging formats. We will use NIfTI:

  • Neuroimaging Informatics Technology Initiative (https://nifti.nimh.nih.gov/nifti-1)
    • essentially a header and data (binary format)
    • will have extension .nii (uncompressed) or .nii.gz (compressed)
    • we will use 3-D images (4-D and 5-D are possible)
  • ANALYZE 7.5 was a precursor to NIfTI
    • had a hdr file (header) and img file (data)

Course Website/Materials

The Course overview is located at (with slides): http://johnmuschelli.com/ISBI_2017.html

All materials for this course (including source for the slides) is located at:


RStudio Server

For this course, we will use an RStudio Server because installing all the packages can be a lengthy process. Mostly all the code we show requires a Linux/Mac OSX platform for FSL and other systems:


Virtual Machine

If you want to run things locally, we have a Virtual Machine you can download and install: https://neuroconductor.org/neuroc-vm

Introduction to R

What is R?

  • R is a language and environment for statistical computing and graphics

  • R was implemented over the S language, which was developed by Bell laboratories

  • R is both open source and open development

(source: http://www.r-project.org/)

Why R?

  • Powerful and flexible

  • Free (open source)

  • Extensive add-on software (packages)

  • Designed for statistical computing

  • High level language

Why not R?

  • Fairly steep learning curve

  • Little centralized support, relies on online community and package developers

  • Slower, and more memory intensive, than the more traditional programming languages (C, Java, Perl, Python)

What comes with R

  • R revolves around functions
    • Commands that take input, performs computations, and returns results
    • When you download R, it has a “base” set of functions/packages (base R)
  • Functions are enclosed in packages
    • A package is collection of functions, documentation, data, and tutorials (called vignettes).
    • Written by R users/developers (like us) - some are bad
    • You install a package using the install.packages command/function:

install.packages is a function, "oro.nifti" is a character string.


Loading Packages

When you install a package, it’s downloaded to the hard disk. That doesn’t mean that you can use the functions from that package just yet.

  • You “load”/import a package into memory using the library command

For example, to load the oro.nifti package:


Now, functions from the oro.nifti package can be used.

RStudio (the software)

RStudio is an Integrated Development Environment (IDE) for R (made by RStudio the company)

  • It helps the user effectively and more easily use R.
    • Syntax highlighting, code completion, and smart indentation
    • Easily manage multiple working directories and projects
  • Is NOT dropdown statistical tools (such as Stata)

R essentially is a command line with a set of functions loaded

RStudio/R Console

  • Where code is executed (where things happen)
  • You can type here for things interactively
  • Code is not saved on your disk
    • Can act as a calculator
    • Creating variables/objects
    • Applying functions

RStudio/R Console