Statistical Methods for Neuroimaging

Goals

• Goal 1: Provide basic knowledge of neuroimaging so that you (Biostatisticians) can read and understand neuroimaging research.
• Goal 2: Cover modern contributions of biostatistics to neuroimaging methods so you can contribute to novel research in the field.

Description

This course covers standard and advanced methods for neuroimage analysis from a biostatistical perspective. Students will learn to analyze and interpret common modalities such as fMRI, structural MRI, cortical thickness, diffusion-weighted imaging, and resting-state connectivity using popular neuroimaging analysis software and visualization tools. Advanced topics may include site/scanner correction, first-level and group-level models, network analysis, AI/machine learning, circularity analysis, multivariate/spatial inference, confidence set methods, and centile methods. Upon completion, students will be prepared to understand and contribute to statistical research in neuroimaging.

Learning objectives

• Gain a foundational understanding of key neuroimaging modalities and their applications in research.
• Develop practical skills for preprocessing neuroimaging data using FSL.
• Apply statistical analysis techniques to neuroimaging data, including first-level, group-level, and functional connectivity analyses.
• Understand/apply modern topics in neuroimaging data analysis such as site/scanner effects, circularity, multiple testing, machine learning, and Bayesian approaches.

Context

• Who are you (the student): Graduate students with a background in biostatistics or related fields.
• Prerequisites: Prior coursework in statistics or biostatistics and experience with R or Python (I will use R).

Grading (Tentative)

Your grade will be determined from

• Quizzes (25%): These will be basic questions about terminology and course content. Grading is based on correct/incorrect answers.
• Mini projects (50%): These will be smaller hands on projects to get you basic exposure to the data and analysis pipelines. We will use a rubric with instructor (me)/peer (you)/ChatGPT (AI) review. TBD.
• Final project (25%): These will be a bigger hands on project to get you basic exposure to the data and analysis pipelines. Probably in groups. We will use a rubric with instructor (me)/peer (you)/ChatGPT (AI) review. TBD.

Links

• Brightspace - For grading and quizzes.
• Course box directory - For all course materials, including these
• Datasets
• Images used in examples in-class

Potential Topics (Tentative)

Introduction

• Introduction and setup
  • Course Introduction
  • Accessing repository data, Data Use Agreements (DUA)
  • Installing FSL

Data Modalities

• Introduction to Neuroimaging Modalities
  • fMRI and rs-fMRI Overview
  • DTI
  • Structural Imaging

Preprocessing and First-Level Analysis

• Preprocessing Pipeline
  • Structural
  • fMRI, rs-fMRI
• Preprocessing with FSL: Practical Session
• Preprocessing: Brainlife.io/Docker
• First-Level Statistical Analysis
  • fMRI, rs-fMRI Analysis
• Practical implementation

Group-Level Analysis

• Multiplicity, Type 1 error, FWER, FDR
• Cluster extent inference, Gaussian field theory, TFCE, Excursion set inference
• Functional Connectivity: basic group-level analysis
• In-class activity: implementation in FSL and R
• “Voodoo” correlation
• Advanced group-level analysis (distance-based)
  • MDMR, PermANOVA, Semiparametrics

Modern Topics: Reproducibility and Replicability

• Reproducibility
  • “The Garden of Forking Paths”
• Replicability
  • Sample sizes
  • Effect sizes

Modern Topics: Machine Learning in Neuroimaging

• Brain-behavior associations (e.g. “Multivariate BWAS”)
  • Prediction
  • Statistical inference
• Replicability in ML Methods
  • Circularity, Data Leakage, and Feature Selection
  • Cross-validation, Bootstrapping
• Guest lecture from Megan?
• Brain Age
  • In-class activity
• Other topics
  • MVPA
  • ML methods
  • Deep learning

Modern Topics: Batch Effects

• Illustration with data so far
• ComBat
• CovBat
• Deep learning

Modern Topics: Misc

• Centile Analysis
• Multimodal Image Analysis

Course Schedule

Code

library(knitr)
library(kableExtra)

# Read the CSV file
schedule = read.csv("courseSchedule.csv", na.strings="NA", check.names=FALSE)

# Display nicely
 kable(schedule, caption = "Neuroimaging Course Schedule", align = "l", escape=FALSE)

Neuroimaging Course Schedule

Date	Day	Section	Topic	In class interactive	Assignment	Due date
20-Aug	Wednesday	Introduction	Introduction and Setup	Software debug	Quiz & install software	27-Aug
25-Aug	Monday	Data Modalities	Introduction to Neuroimaging Modalities
27-Aug	Wednesday	Preprocessing	Preprocessing Pipeline	Registration with FSL	Quiz	12-Sep
1-Sep	Monday		No class: Labor day
3-Sep	Wednesday	Preprocessing	Structural preprocessing
8-Sep	Monday	Preprocessing	Recorded: Structural and functional preprocessing		Despicable Me preprocessing	22-Sep
10-Sep	Wednesday		Recorded: Functional preprocessing	Recorded: Despicable Me preprocessing
15-Sep	Monday	First-Level Analysis	First-Level fMRI Analysis		Quiz	26-Sep
17-Sep	Wednesday	First-Level Analysis	Time Series Analysis		First-level fMRI analysis of Despicable Me data	6-Oct
22-Sep	Monday	First-Level Analysis	rs-fMRI First-Level Analysis	Recorded: Despicable Me first-level analysis
24-Sep	Wednesday	Group-Level Analysis	Recorded: Estimation		Quiz- see Brightspace	30-Sep
29-Sep	Monday	Group-Level Analysis	Recorded: Multiplicity, Type 1 error, FWER, FDR
1-Oct	Wednesday	Group-Level Analysis		Group-level ABIDE analysis
6-Oct	Monday	Group-Level Analysis		Group-level ABIDE analysis
8-Oct	Wednesday	Group-Level Analysis
13-Oct	Monday	Group-Level Analysis	Cluster extent inference, Gaussian field theory, TFCE, Excursion set inference		Group-level analysis of Despicable Me	31-Oct
15-Oct	Wednesday	Group-Level Analysis		Group-level ABIDE analysis
20-Oct	Monday	Common Mistakes and Ongoing Challenges	“Voodoo” correlation, circularity errors, data leakage, the salmon	Group-level ABIDE analysis
22-Oct	Wednesday	Modern Topics	Replicability, Inference for ML accuracy (Megan Jones)
27-Oct	Monday	Modern Topics	Confidence sets (Xinyu Zhang)
29-Oct	Wednesday	Modern Topics	Overview of projects		Draft paper (for peer review)	1-Dec
3-Nov	Monday			Working on projects
5-Nov	Wednesday			Working on projects
10-Nov	Monday			Working on projects
12-Nov	Wednesday		Common mistakes and challenges
17-Nov	Monday			Working on projects
19-Nov	Wednesday			Interim project presentations
24-Nov	Monday		No class: Thanksgiving break
26-Nov	Wednesday		No class: Thanksgiving break
1-Dec	Monday		Peer reviewing papers		Final paper (per group)	5-Dec
3-Dec	Wednesday		Final project presentations