In the last ten years, the tools of machine learning have been introduced in the field of in vivo brain functional exploration methodologies to identify distinctive features that can allow “brain-reading”, prediction of mental states or behavior, or recognition of mental disorders directly from brain functional data. Indeed, brain functional imaging has long been used to test specific hypotheses about brain-behavior relationships, generally by noting that previous studies have found an area to be engaged for a particular mental process and inferring that this process must be engaged whenever that region is found to be active. However, recent developments in the application of statistical classifiers to brain functional data provide the means to directly test how accurately mental processes or cognitive states can be classified.During my lecture, I will present different decoding and encoding approaches, developed in our laboratory, to fMRI analysis using powerful pattern-classification algorithms to decode the information that is represented in that pattern of activity. In particular, machine learning approaches have been applied to multi-voxel patterns of functional activity to assess processing of sensory information, to decode semantic representation and to encode/decode movements. These novel approaches are not only providing original information to understand brain functioning, but also enhancing the advantages of brain functional imaging for clinical neuroscience, neurorehabilitation and bioengineering.
|Titolo:||Decoding and encoding approaches to brain imaging data: from cognition to robotics|
|Data di pubblicazione:||2016|
|Appare nelle tipologie:||4.1 Contributo in Atti di convegno|