Information

The estimated time to complete this training module is 2h.

The prerequisites to take this module are:

• the installation module.
• the introduction to python for data analysis module.

If you have any questions regarding the module content please ask them in the relevant module channel on the school Discord server. If you do not have access to the server and would like to join, please send us an email at school [dot] brainhack [at] gmail [dot] com.

Resources

This module will be using the MAIN educational workshop on brain encoding and decoding.

Exercise

Let’s have a look at application on functional data with the MAIN educational workshop on brain encoding and decoding. We will look at the decoding modules. This part is independent from the video above.

• Please follow the introduction, set-up your environment and clone the material from GitHub. Please provide your answers in jupyter notebooks.

• Throughout this tutorial you will be using the Haxby data set. Please read through and understand how to access it here and go through the original support-vector machine analysis of the study and complete the exercises.

• After understanding the workflow of functional data, please go through the Multi-Layer Perceptron and complete the relevant exercise. If you want a challenge, please feel free to do the harder questions, or do both lessons 🎉.

• (Optional) You can learn about the Graph Convolution Network and how to work with timeseries data!

• Follow up with your local TA(s) to validate you completed the exercises correctly.

• 🎉 🎉 🎉 you completed this training module! 🎉 🎉 🎉

More resources

This demo was presented by Jakub Kaczmarzyk during the QLSC 612 course in 2020, the slides are available here.

The video of the presentation is available below:

You can find more information about the nobrainer library on its github repo.

A nature communications article on the superiority of deep learning over standard machine learning in neuromimaging tasks.

A Neuroscience and Biobehavioral Reviews article on deep learnging applications in neuroimaging studies of brain-based disorders. It has a good overview of the general framework of deep learning applications, and descriptions of the main kinds of architectures.

An overview article on the challenges associated with applying deep learning models to neuroimaging applications, especially fMRI data. Three new methods are presented which could potentially help incorporate these models into clinical practice.

MAIN educational workshop on brain encoding and decoding covers deep learning application to analyse a classic neuroimaging dataset. The tutorial also incorporates useful features from the nilearn library to process your neuroimaging data, as well as doing decoding analysis.