Elizabeth Sweeney

Elizabeth Sweeney

Assistant Professor of Biostatistics

Weill Cornell

I am an Assistant Professor in the Biostatistics Division in the Population Health Sciences Department at Weill Cornell Medicine. My main area of expertise is working on the statistical analysis of structural magnetic resonance imaging (sMRI) data, with a focus on the disease area of Multiple Sclerosis (MS). The problems I have worked on span many areas, including image segmentation, image normalization and harmonization, cross-sectional and longitudinal modeling, as well as software development. The statistical techniques used to solve these problems include classification, machine learning, longitudinal data analysis, and functional data analysis techniques. In addition, I am an avid R enthusiast! I am currently an associate editor for the R Journal and served on the board of the meetup group R Ladies NYC. A video of my 2019 NYR talk on doing neuroimaging analysis in R can be found here. I also co-taught a Coursera course on neuroimaging analysis using R called Introduction to Neurohacking in R. You can contact me at ems4003@med.cornell.edu.

Download my CV.

Interests
  • Structural Magnetic Resonance Imaging
  • Multiple Sclerosis
  • Biostatistics
Education
  • PhD in Biostatistics, 2016

    The Johns Hopkins Bloomberg School of Public Health

  • ScM in Biostatistics, 2012

    The Johns Hopkins Bloomberg School of Public Health

  • BSc in Mathematics, 2010

    Purdue University, Indianapolis

Experience

 
 
 
 
 
Assistant Professor
May 2019 – Present New York
 
 
 
 
 
Senior Data Scientist
May 2018 – Apr 2019 New York
 
 
 
 
 
Quantitative Scientist/ Senior Quantitative Scientist
Feb 2017 – May 2018 New York

Projects

.js-id-QSM
Quantitative Susceptibility Mapping
In collaboration with Susan Gauthier, Thanh Nguyen, Yi Wang, and Sandra Hurtado Rua we are working on methods to analyze the behavior of multiple sclerosis lesions on a structural magnetic resonance imaging (sMRI) sequence called the quantitative susceptibility map (QSM). QSM is a new sMRI technique that provides in-vivo quantification of magnetic susceptibility changes related to iron deposition. We are interested in a subtype of lesion with chronic inflammation, which we call QSM rim+ lesion. QSM rim+ lesion have a distinct longitudinal pattern consistent with retained inflammation followed by a transition to a chronic inactive state. Weare actively working to understand how demographic features and treatment impact the longitudinal behavior of these lesions.
Quantitative Susceptibility Mapping
Multiple Sclerosis Lesion Age Estimation
When a multiple sclerosis (MS) lesion is detected in the brain using structural magnetic resonance imaging (sMRI) it is often unknown how old this lesion is. Information about the age of an MS lesion is crucial for analyzing longitudinal lesion information as well as for diagnosing the disease. We developed methodology to estimate the age of an MS lesion from cross-sectional multisequence sMRI studies using lesion radiomic features (histogram based, shape based, and texture based features) (publication). Currently, we are expanding upon this work to utilize longitudinal information from MS lesions for more accurate age estimation. We are implementing functional data analysis (fda) techniques to register multisequence longitudinal curves from MS lesion to estimate this age.
Multiple Sclerosis Lesion Age Estimation
Longitudinal Multiple Sclerosis Lesion Behavior
Brain structural magnetic resonance imaging (sMRI) is a tool that uses a magnetic field to produce detailed images of the brain. Patients with multiple sclerosis (MS) have lesions in their brains which are visible on sMRI. MS lesions formation is a complex process involving inflammation, tissue damage, and tissue repair — all of which are visible on sMRI and potentially modifiable by pharmacological therapy. We developed a piepline to extract longitudinal information from MS lesions in multisequence sMRI. We then developed a biomarker of lesion repair and explored the biomarker’s association to clinical information, such as the use of steroids and disease-modifying treatment (publication).
Longitudinal Multiple Sclerosis Lesion Behavior
Multiple Sclerosis Lesion Segmentation
Magnetic resonance imaging (MRI) can be used to detect lesions in the brains of multiple sclerosis (MS) patients and is essential for diagnosing the disease and monitoring its progression. In practice, lesion load is often quantified by either manual or semi-automated segmentation of MRI, which is time-consuming, costly, and associated with large inter- and intra-observer variability. We developed the method OASIS for automated lesion segmentation at a single time point (publication) and SuBLIME for segmentation of incident MS lesions (publication). We also performed a comparison of different machine learning algorithm for the problem of MS lesion segmentation (publication). More recently we have turned our attention to identifying a particular subtype of lesion, called a chronic active MS lesions, which has ongoing inflammation at the edges of the lesion. These lesions are of interest because they have the potential to serve as a biomarker in MS and may even help to evaluate disease-modifying treatments. These lesions are characterized by a hyperintense rim on an MRI sequence called quantitative susceptibility mapping (QSM) and a hypointense rim on phase imaging. We have worked on methods to identify these lesions on both phase (publication) and QSM imaging.
Multiple Sclerosis Lesion Segmentation

Teaching

.js-id-ds1
Data Science I (R and Python)
Data Science I provides an introduction to data science using both the R and python programming languages. In this course students gain experience working directly with data to pose and answer questions. The course is divided into two parts; the first part is taught with the programming language R and the second with python. Topics covered include: reproducible research, exploratory data analysis, data manipulation, data visualization techniques, simulation design, and unsupervised learning methods. Taught at Weill Cornell Medicine from 2019 to present.
Data Science I (R and Python)
Longitudinal Data Analysis
Longitudinal data analysis focuses on methods for the analysis of longitudinal data, with an emphasis on how to perform these analyses in R. The course covers basic exploratory data analysis for longitudinal data, marginal and random effects models for both continuous and binary outcomes, and other special topics in longitudinal analysis. Taught at Columbia University in 2018.
Longitudinal Data Analysis
Introduction to Neurohacking in R
Neurohacking describes how to use the R programming language and its associated packages to perform manipulation, processing, and analysis of neuroimaging data. We focus on publicly-available structural magnetic resonance imaging (sMRI) datasets. We discuss concepts such as inhomogeneity correction, image registration, and image visualization. The course is ongoing on Coursera.

Publications

(2021). Estimation of Multiple Sclerosis lesion age on magnetic resonance imaging. Neuroimage.

Cite

(2021). Increasing age is independently associated with higher free water in non-active MS brain-A multi-compartment analysis using FAST-T2. bioRxiv.

Cite

(2020). A Consensus Time for Performing Quality Control of 225Ac-Labeled Radiopharmaceuticals.

Cite

(2020). Fully Automated Detection of Paramagnetic Rims in Multiple Sclerosis Lesions on 3T Susceptibility-Based MR Imaging. bioRxiv.

Cite

(2020). Quantifying cognitive resilience in Alzheimer?s Disease: The Alzheimer?s Disease Cognitive Resilience Score. PloS one.

Cite