Principal Investigators //

Sergey Plis, PhD

Assistant Professor of Translational Neuroscience, Director of Machine Learning in Neuroscience Lab

Sergey Plis

Dr. Plis researches novel techniques and approaches to analyzing multimodal brain imaging datasets. The main tool source is the field of machine learning. The main goal is to be able to infer structures and patterns in brain function that are hard to obtain non-invasively and/or are unavailable for direct observation. In the long term, this develops methods able to teach us about mechanisms used by the brain for forming task specific transient interaction networks. Dr. Plis' current research 
is on inferring probabilistic descriptions of these function induced networks based on fusion of fast and slow imaging modalities: MEG and fMRI. 

Email Dr. Plis

Capturing Complex Interactions in Neuroimaging Data

Unsupervised data analysis approaches have been widely used in recent years  and  have  become  instrumental in  establishing  new  research directions  impossible with  more  traditional supervised  approaches, such as the  study of the default mode network  of the brain.  Growing interest in unsupervised analysis of multi-modal data, multi-subject studies, whole brain activit and other datasets  involving multiple interacting   variables,  have   increased  demand   for  multivariate unsupervised  techniques.  However there is still relatively little work on the examination of  the full relationships  among interacting variables in an  unsupervised fashion.  In this project, we work on a novel   approach to   address the   problem  of   identifying  these higher-order  interactions. Unlike existing approaches, we directly identify sets  of inter-dependent random  variables without explicitly modeling interactions within these sets.

Multimodal Data Fusion for Brain Connectivity Analysis

Despite enormous strides in  our understanding of  neural physiology, the transition from cellular and subcellular functional variability to variations in  human behavior is poorly  understood.  Current evidence ties this transition to complex interactions  within brain functional networks. These networks are not easy to estimate from the data due to their dynamic  nature. However,  accurately characterizing them  is of the  uttermost  importance  for  diagnosis and  prediction  of  mental disorders at their early stages (e.g. schizophrenia).  This project is aiming to  improve estimation of  brain functional networks  by taking advantage  of  the  complementary  nature of  multiple  brain  imaging modalities. The main  application is in the study  of mental disorders linked to brain network dysfunction (schizophrenia, Alzheimer's, ADHD, bipolar disorder and others).