Statistical Methods for Medical Image Analysis

Tutorial Instructors:


Dr. Davatzikos is the Wallace T. Miller Sr. Professor of Radiology at the University of Pennsylvania. He is also Director of the Center for Biomedical Image Computing Analysis and Director of the Section of Biomedical Image Analysis at Penn. He was named a Fellow of the Institute of Electrical and Electronics Engineers (IEEE) in 2014.


Dr. Linn is an Assistant Professor of Biostatistics at Penn. She earned her PhD in statistics from North Carolina State University. Her research focuses on adapting and applying statistical methods to multimodal brain imaging. Her work incorporates ideas from causal inference and machine learning to reduce the effects of confounding and improve the interpretability and generalizability of estimated multivariate disease patterns.


Dr. Shinohara is an Assistant Professor of Biostatistics at Penn. He earned his PhD in Biostatistics from the Johns Hopkins University. Dr. Shinohara works in neuroimaging statistics, high-dimensional data, and causal inference. His laboratory focuses on statistical methods for biomedical imaging with applications in multiple sclerosis, Alzheimer’s disease, depression, post-traumatic stress disorder, autism spectrum disorders, and brain development.


Dr. Shou joined Penn as an Assistant Professor of Biostatistics after completing her PhD in Biostatistics at the Johns Hopkins University. Her research focuses on statistical methods for high-dimensional data with complex temporal or spatial structures, particularly multimodal imaging from mental health studies. She has contributed methods to quantify variability and extract features from MRI and diffusion tensor imaging, assess reproducibility and correct measurement error for functional connectivity maps, and predict disease outcomes using imaging markers.


Dr. Vandekar will be joining the Department of Biostatistics at Vanderbilt University in Summer 2018 as an Assistant Professor. He earned his PhD in Biostatistics from the University of Pennsylvania. Dr. Vandekar has developed techniques for multi-modal imaging data analysis and statistical methods that address issues of multiple hypothesis testing with neuroimaging data. He has applied these methods to study neurodevelopment and Alzheimer’s disease.

Tentative Schedule:

Module 1. Statistical Assessment of Scan-Rescan Reliability (Shou)

Module 2. Multi-scanner Harmonization of Imaging Data and Replicability Analysis (Shinohara)


Module 3. Multiple Comparison Correction (Vandekar)

Module 4. Confounding and Multivariate Pattern Analysis (Linn)

Module 5. Statistical Significance Maps for Machine Learning Methods (Davatzikos)