Machine Learning in Medical Imaging

Project description: Recently, there has been a significant progress on the application of machine learning approaches to neuroimaging analysis. These approaches focus on predicting a variable of interest (e.g. patients vs. healthy subjects) based on the pattern of brain activation/anatomy over a set of voxels. Due to their multivariate properties, these methods can achieve relatively greater sensitivity and are therefore able to detect subtle, spatially distributed activations and patterns of brain anatomy. In addition, the predictive framework of machine learning methods enables its application to new data which is very relevant for clinical applications as they can be used as diagnostic/prognostic approaches. This project will introduce participants to some of the machine learning techniques currently used to analyse neuroimaging data, using the PRoNTo toolbox.

Associated publications:

http://dx.doi.org/10.1007/s12021-013-9178-1

https://link.springer.com/article/10.1007/s12021-017-9347-8

 Prerequisites:

Familiarity with basic neuroimaging concepts. No prior machine learning knowledge is required. Participants should bring a laptop with MATLAB installed (currently, versions 7.10 (R2010a) to 9.4 (R2018a) are supported by PRoNTo), and SPM (12 with latest updates). It also requires the MATLAB statistics toolbox.