Intro to Analyzing Brain Imaging Data— Part I: fMRI Data Structure
There is a growing interest in applying machine learning techniques on medical data. Brain scans from Magnetic Resonance Imaging experiments (MRI) have been a popular choice with the number of publications combining MRI and machine learning growing exponentially over the last years (see data from PubMed below). Therefore in this first post we will cover some of the basics about structural and functional MRI (fMRI) data to give you an idea of how the data is generally structured. In the following post we will analyze the data by doing some correlation analysis and by building a general linear model (GLM) to identify active regions in the brain.
The focus of these posts will be on the structure and analysis of the data and not on the underlying principles of magnetic resonance imaging.
Structural MRI images
Structural MRI scans usually visualize the location of water in the human body. This means that soft tissues with high water and fat concentration such as the brain can be well resolved while more dense structures such as bones have a lower signal amplitude. Structural MRI scans allow clinicians to…
