The R programming language: a statistical foundation for reproducible studies in medical image analysis
Please also clone https://github.com/stnava/RMI_Data with the RMI/stnava directory
Outline: This 1/2 day tutorial introduces attendees to the most widely used and deeply tested statistical platform in the world and will reveal the secrets to using R for reproducible science and advanced statistics in medical imaging applications. The R programming language is increasingly employed for communicating scientific results. Eminent journals such as Biostatistics promote articles that include R code reproducing the submission’s findings. Industry also embraces R for its standard-setting analytics. Recent efforts allow R to function efficiently with medical imaging datasets. R therefore allows medical imaging researchers access to state-of-the-art methods developed by the world’s leading statisticians. This tutorial will show how, with relative ease, attendees can process medical imaging datasets in a reproducible way. This MRI-focused tutorial will step attendees through examples that range from basic I/O, data inspection and study organization to morphometry, fMRI connectivity analysis, DCEMRI and longitudinal statistics.
Primary organizer: Brian B. Avants Assistant Professor, Department of Radiology, University of Pennsylvania Email: [email protected] 3600 Market St., Suite 370, Philadelphia, PA, 19104 brianavants.wordpress.com
Co-organizer: Tom Fletcher, Assistant Professor, School of Computing, University of Utah Email: [email protected] Warnock Engineering Building 72 South Campus Central Dr., Room 3750 University of Utah Salt Lake City, Utah 84112 www.sci.utah.edu/~fletcher/
The primary objective of the tutorial is to demystify and democratize R usage within medical imaging. There is significant interest in this software platform and successful imaging-focused libraries for R are very recent. The platform may be used to address numerous emerging statistical problems within the medical imaging community, in particular imaging genomics, structural and fMRI network analysis (package ANTsR), medical imaging data formats (packages oro.dicom amd oro.nifti), DCE-MRI (package dcemriS4) and longitudinal analysis (package lme4). In addition, the fundamentals of statistics are trivially available such as linear regression, ANCOVA and mixed-effects modeling. Topics will also include visualization, model selection and exploration of high-dimensional, heterogeneous datasets common in imaging studies that involve multiple imaging modalities and both categorical and continuous clinical variables. We also anticipate that this inaugural tutorial will grow in importance in the future and will be revisited several times with focus changing according to contemporaneous interests.
The tutorial will consist of 3 presentations with ample time for discussion and questions. Presentations will consist of background material along with roughly 1/2 of the time focused on reproducible examples. The tutorial will suggest that attendees install relevant software prior to the meeting although time will be available to assist those for whom this was not possible. The material generated for this tutorial will also serve attendees off-site and post-tutorial as a high-level and practical introduction to medical image processing with R.