CONN Toolbox Workshop 2020

Welcome to the OSU CONN toolbox workshop! To prepare for the course, do the following steps:

1. Get Started with Matlab

Please watch this playlist for an introduction to Matlab and its basic commands. It is also recommended that you download version R2017a or later.

2. Install the CONN Toolbox

Use this link for instructions on how to download and install the CONN Toolbox.

Note that CONN also requires the SPM package; instructions for downloading and installing SPM can be found here.

3. Download the Dataset

The dataset we will use for this workshop is from openneuro, which can be found here.

You can also download a sample functional connectivity image here.

Day 1: fMRI Pitfalls

func_conn_maps.jpg

Agenda

(9:30am-11:00am) Overview of Issues in fMRI Analysis (Lecture)
How have fMRI analysis problems changed over time? This lecture looks at the history of controversial issues that have come up.

  • The structure of fMRI data: Time-series, signal intensity, and how artifacts are introduced

  • Dead salmon and multiple comparisons: How to correct for multiple tests and why

  • Voodoo correlations: How to spot biased region of interest (ROI) analyses and how to avoid doing them yourself

  • Cluster failure: How to accurately calculate your cluster threshold

  • Reproducibility: Tools for making your data easier to analyze by other groups. Overview of openneuro, BIDS, and other online repositories

(11:00am-12:00pm) How Not to Do It (Practical)
One of the best ways to spot questionable research practices is to be aware of when you do them yourself. (The sample dataset can be found here.) This practical will guide you through some of the most common scenarios, including:

  • Creating your own biased analysis: circular analyses using your own data, and data peeking (using this script)

  • Cluster thresholds

  • Orthogonalization of task-based regressors and parametric modulators

  • QC checks to guard against spurious results

    • Checking timing files through reverse inference

    • Correlated regressors and what to do about them

(12:00pm-1:00pm) LUNCH BREAK

(1:00pm-2:00pm) Pitfalls of Functional Connectivity (Lecture)
A brief overview of functional connectivity, and some of the common pitfalls encountered in both task-based and resting-state connectivity analyses

  • Functional vs. effective connectivity

  • Basics of functional connectivity: Scrubbing, ROIs, correlations between regions

  • Resting-state specific preprocessing

  • The debate over Global Signal Regression

  • Motion artifact removal since 2011: What are the best approaches?

(2:00pm-2:45pm) Introduction to the CONN Toolbox (Lecture & Practical)
After reviewing some of the most common errors in both task-based and resting-state analysis, we will begin working with the CONN toolbox: a software package for connectivity analysis. During this session we will double-check our installation of CONN and related toolboxes (e.g., ART) and take a tour through the graphical user interface.

(2:45-3:30pm) Preprocessing the Individual Subject (Lecture & Practical)
This practical will show how to preprocess the data for a single subject in CONN.

  • Realignment, slice-timing, and outlier detection

  • Loading the SPM.mat file

  • Data denoising and detrending

  • Checking normalization and smoothing

(3:30pm) Photo Shoot
We will take a group photo at the end of the first day.

(3:30pm-5:00pm) Individual Consulting Sessions
Andy will help individuals (or small groups of individuals) with their data. One session runs from 3:30pm-4:15pm, and the second one runs from 4:15pm-5:00pm. These meetings will be set up ahead of the workshop.

Day 2: Functional Connectivity: Practice with the CONN Toolbox and Advanced Options

CONN_Image.jpg

Agenda

(9:00am-10:00am) Quality Assurance (Practical)
As with all neuroimaging data, quality assurance checks are very important. After reviewing some common examples of quality failures, you will examine your own data.

  • Checking registration between T1 and T2 modalities

  • Alignment of the outline of the brain vs. alignment of internal structures

  • Outlier counts and detection

(10:00am-10:45am) Group-level analysis (Lecture & Practical)
An overview of how to set up group-level analyses, as well as caveats to be aware of.

  • Setting up group-level analyses in CONN

  • Second-level covariates and between-subject factors

  • Correction mechanisms: FWE, FDR, and cluster-forming thresholds

(10:45am-11:30am) Viewing the Results (Practical)
We will tour CONN’s visualization tools, and learn how to interpret the results in light of the graph theory terms we just learned.

  • Thresholding your connectivity maps

  • Rendering connectivity maps onto surfaces

  • Viewing node connectivity maps

In addition, we will do a brief overview of graph theory. Graph theory is closely related to the correlations between different brain regions. This second lecture will cover the fundamentals you will need to understand the basic terms.

  • Background of graph theory

  • Basics of graph theory: Nodes, edges, modularity,

  • Community detection

  • Correlation matrices and clinical applications

(11:30am-12:00pm) Setting up a PsychoPhysiological Interaction (PPI)
The COON toolbox is also able to do task-based connectivity through a method known as PPI. We will demonstrate how to set up a generalized PPI (gPPI) analysis for a single subject. A custom atlas and covariate files can be found here.

  • Overview of PPIs

  • Importing timing files

  • Importing atlases

(12:00pm-1:00pm) LUNCH BREAK

(1:00pm-1:30pm) Scripting your analysis
Automating analyses is an indispensable skill for the neuroimaging researcher. This practical will demonstrate how to script analyses in the CONN toolbox, which can speed up processing and reduce the likelihood of error. A template script can be downloaded here.

  • Creating your analysis script

  • Looping your analysis over subjects

  • How to obtain converted correlation-to-z values in a Matlab structure

(1:30pm-3:00pm) Analyzing a New Dataset (Practical)
The concepts and tools we have discussed so far should provide you with enough knowledge to begin to analyze your own data. This session will be devoted to beginning your own analysis with CONN. This can either be your own data, or a dataset downloaded from an online repository, such as humanconnectome.org or openneuro.org.
As a group, we will select an open-access dataset to analyze (options will be provided by Andy). We will then determine how to analyze the dataset, and begin the processing of a single subject.

(3:00pm-3:45pm) Individual Consulting Sessions
Andy will help individuals (or small groups of individuals) with their data, from 3:00pm-3:45pm.