Advanced neuroimaging research for the digital age
Bringing AI and Neuroimaging together to transform medicine
My Interests
I am a Research Assistant Professor at Northwestern University’s Feinberg School of Medicine in Chicago, IL, USA.
My interest is understanding the basis of structural and functional deficits in patients with neurodegenerative diseases such as Parkinson’s disease. In order to identify changes in brain morphology and function in patients, I develop advanced structural, microstructural and functional MRI analytic techniques. I use these methods to improve early detection, differential diagnosis, tracking of longitudinal changes and evaluating treatment interventions.
In order to improve workflow and enable robust, repeatable data analyses, we integrate these advanced neuroimaging pipelines into the Northwestern University Research Image Processing System (NURIPS). I am Co-Director of the NURIPS system. NURIPS is the next generation image data repository and processing system that replaces the retired Northwestern University Neuroimaging Data Archive (NUNDA) system. This web-based database system stores raw imaging scans, transfers data seamlessly to multiple high performance computing environments such as Quest, processes scans through version-controlled automated imaging pipelines and stores results back with the original subject scans. This system enables robust, automated workflows that take advantage of HPC resources and is designed to integrate with next-gen deep learning workflows for model training and deployment.
Finally, I am of the core faculty members of the Augmented Intelligence in Medical Imaging working group that is part of the recently formed Institute for Augmented Intelligence in Medicine at Northwestern University. The goal is to develop deep learning models that can inform us of which brain regions are driving patient classification, along with developing predictive models that can use MRI inputs to predict future decline, improve image processing tasks such as correcting motion artifacts and image segmentation..
NURIPS
The Northwestern University Research Image Processing System (NURIPS) both archives and processes imaging data via a web-based interface. As Co-Director of this system, my mission is to bring the next generation of automated workflows and cloud computing capabilities to imaging research at Northwestern University.
MULTIMODAL MRI IN NEURODEGENERATIVE DISEASES
In order to gain insight into underlying pathways and mechanisms impacted by neurodegenerative diseases such as Parkinson’s disease and Alzheimer’s disease, I leverage multimodal microstructural, structural and functional imaging. To improve reliability of these algorithms I have adopted practices common in industry such as GIT for version control of code, markdown language for clear code documentation and singularity for running production ready code on multiple systems while maintaining data consistency. My custom pipelines are optimized for accuracy in neurodegenerative disease populations in order to improve sensitivity and specificity of results.
My goal is to then integrate our advanced preprocessing methods with NURIPS, which utilizes high performance computing resources to tackle data preprocessing and analysis at larger scales. This will enable faster development of deep learning algorithms that can aid in differential diagnosis, deep learning predictive models of motor and cognitive decline, along with identifying which regions of the brain are driving the outputs of these models in order to further explore the biological underpinnings.
Augmented intelligence
As a core member of the Augmented Intelligence in Medical Imaging (AIMI) working group in the newly formed Institute of Augmented Intelligence in Medicine at Northwestern University, the focus of our group is to apply augmented intelligence (AI) to create transformative medical AI-based applications to help clinicians improve patient outcomes through personalized care across all medical imaging disciplines and diseases. My interest is to develop and apply cutting-edge deep learning algorithms to create predictive models of motor and cognitive decline in Parkinson’s disease and atypical Parkinsonism.
