Program
The DeepMI FastSurfer/FreeSurfer course will be divided into several modules that give a comprehensive overview of the FastSurfer and FreeSurfer software packages and provide detailed guidance for their application in your own research. After an introduction, the segmentation module will cover advanced deep-learning methods for image segmentation. The surface module will explain how 3D surface models are created from brain images. The group analysis module will discuss statistical modeling and inference, and the QC and Add-ons module provides QC and troubleshooting tips as well as an outlook into the current and future FastSurfer ecosystem. All lectures will be accompanied by demos and/or practical exercises.
Introduction to structural neuroimaging
This module provides an overview of the FastSurfer / FreeSurfer software packages, their image processing pipelines, and the resulting metrics for quantifying anatomical properties of the human brain. This includes:
- A brief introduction to artificial intelligence and human neuroimaging research
- Application of FastSurfer and FreeSurfer in human neuroimaging studies
- Properties, commonalities and differences of the FastSurfer and FreeSurfer software packages
This module also includes a short tutorial on working with our Linux workstations and on running scripts via the command line.
Advanced deep-learning for image segmentation
This session covers the core module of the FastSurfer software – fast and reliable anatomical segmentation. We will focus on the analysis of individual cases and discuss the following topics:
- Deep-learning-based segmentation of anatomical brain images
- From images to metrics: quantification of anatomical information
- Pipelines, commands, outputs: practical and technical aspects
- How to get everything up and running: installation, setup, and deployment of the FastSurfer and FreeSurfer software
This module includes a practical session during which particpants can conduct their own segmentation analysis using FastSurfer and evaluate the resulting output.
Surface models
This module continues the analysis of individual cases and introduces the creation of surface models and how they can be used to extract a rich set of anatomical features. It will cover the following topics:
- What are surface models - and why should we use them?
- What surface-based outputs can we get - and how can we use them?
- Surface registration and the preparation of statistical analysis
A practical session will allow participants to become familiar with the full FastSurfer / FreeSurfer output as well as the Freeview program, our primary tool for the visualization and inspection of individual results.
Statistical modeling and inference
This module covers statistical modeling and inference. We will address how data from individual cases can be aggregated at the group level for the purpose of answering applied research questions (e.g., group differences, association studies). We will discuss:
- Application of the general linear model in neuroimaging analyses
- The statistical design: hypotheses, contrasts, statistical parameters
- Multiple comparison correction in mass-univariate analyses
- Commands, output files, and visualization using the Freeview software
A practical session covers the complete workflow of an analysis at the group level, using preprocessed data from a prototypical study.
QC, troubleshooting, and practical issues
This module illustrates potential pitfalls and issues during a FastSurfer / FreeSurfer analysis, and provides suggestions for mitigating them. It will cover:
- Quality control: automatic and visual inspection of processing results
- Troubleshooting: what to do in case of processing failures or segmentation errors
- Caveats: scenarios when (and how) to use FastSurfer or FreeSurfer - and when not
During a practical session, participants will learn how to apply strategies and techniques for quality checking. We will also discuss what output can be expected from a FastSurfer/FreeSurfer analysis, and when to proceed with caution.
The FastSurfer ecosystem
The final module introduces the FastSurfer ecosystem, i.e. the set of software tools that extend and smoothly interoperate with the core FastSurfer software. We also give an outlook about future directions in the development of the software. We will cover:
- Specialized application scenarios: infants, lesions, high-resolution or high field-strength data
- Add-on modules for Amygdala/Hippocampus, Thalamus, Hypothalamus, Olfactory Bulb, brainstem, Cerebellum
- Outlook to ongoing and planned developments; interaction with the FastSurfer user community
The course concludes with a question-and-answer session about any potentially remaining issues. There will also be opportunites to meet the speakers individually.