Computational Neuroimaging Laboratories

The Computational Neuroimaging Lab’s research agenda involves the development of novel computational analysis and experimental techniques for determining how brain function and structure are impacted by mental illness and development. Current projects include:

Development of a 9.4T MRI scanner

Together with the Design, Acquisition and Neuromodulation Laboratories (DANL), the Computational Neuroimaging Laboratories (CNL) is leading the efforts for the development and implementation of a 9.4 tesla magnetic resonance imaging system at NKI; it will be one of the most powerful imaging systems in the world for neuroscientists to study the human brain. This neuroimager will be used to safely and noninvasively observe human brain anatomy, metabolism and function at the highest spatial, spectral and temporal resolutions achievable. This instrument will bring together neuroscience researchers from throughout the Greater New York City and Tri-State area. Centrally located in one of world’s top neuroscience communities, this 9.4T neuroimager will foster significant new discovery in understanding the human brain and behavior. Educational and training programs are also supported. A new generation of trained engineers and scientists, and a new level of science and discovery, will follow.

We are developing a novel second-generation console to optimize the quality of data obtained from this ultra-high field strength magnet. This revolutionary engineering project is conducted as a collaboration between NKI, Columbia, NYU and Cornell engineering schools, and New York State industries including GE Global Research and Communications Power Corp. This console will be interfaced to an existing 9.4T head-only magnet to complete the neuro-imaging system. It will be housed in a new ultra-high-field imaging laboratory being built separately for this project. Participating in this project are both academic faculty and industry engineers, students and technicians. This project is funded by the Office of Mental Health of New York State, Research Foundation for Mental Hygiene (RFMH), and a grant from the National Science Foundation (project # 2117823).

Predictive eye estimation regression

The acquisition of eye gaze information during functional magnetic resonance imaging (fMRI) enables the monitoring of variations in attention and task compliance, and such measures are especially important for fMRI paradigms that involve overt behavioral response, such as movie fMRI. This project proposes to extensively enhance a technique called Predictive Eye Estimation Regression (PEER) that uses information from the fMRI signal to predict eye gaze location. This will enable researchers to (re)analyze their fMRI data by using eye gaze information in their models to study brain activity. We will also develop a real-time system to predict eye-gaze location during the fMRI experiment. The system will also track head motion in real-time, thereby increasing data quality and reliability.

Neuroinformatics

The CNL is working in several research projects and is aimed at providing neuroinformatics support. The current projects are:

  • Neurobiology and dynamics of Active Sensing: Data Standardizing and Sharing Core (NIH grant P50MH109429).
  • The NKI Rockland Sample II: An Open Resource of Multimodal Brain, Physiology & Behavior Data from a Community Lifespan Sample (NIH grant R01MH124045).
  • Reproducible imaging-based brain growth charts for psychiatry (NIH grant R01MH120482).
  • International Neuroimaging Data Sharing Initiative (http://fcon_1000.projects.nitrc.org/), with a goal of sharing large neuroimaging datasets with the scientific community.

Quality control of neuroimaging data

The Computational Neuroimaging Laboratory has several previous and current research projects focused on improving the quality and reliability of structural and functional neuroimaging data. We have been working on evaluating the quality of innovative MRI pulse sequences (i.e., the use of prospective motion correction or multi-echo sequences) and preprocessing techniques to reduce physiological noise.