In this section, we took a glimpse at how MRI scanners operate.

As you have seen, MRI scanners are probably some of the most complex inventions made by humankind. The design of one touches on hard problems in the fields of robotics, quantum physics, signal processing, mechanical, electrical, and software engineering.

MRI scanners are very versatile machines that can provide great non-invasive insight into what is happening inside a human body, and they lend themselves to a lot of interesting AI problems.

Further Resources

• Facebook’s research on using AI to accelerate the MRI reconstruction process: Overview and a paper about fastMRI: An Open Dataset and Benchmarks for Accelerated MRI.

• If you want to understand details of MR physics better, here is a paper on the subject, oriented at clinicians: Ridgway, J. P. (2010). Cardiovascular magnetic resonance physics for clinicians: part I. Journal of Cardiovascular Magnetic Resonance, 12(1). doi: 10.1186/1532-429x-12-71.

• If you would like to understand the process of MRI image reconstruction a bit better, here is an excellent in-depth overview: Hansen, M. S., & Kellman, P. (2014). Image reconstruction: An overview for clinicians. Journal of Magnetic Resonance Imaging, 41(3), 573–585. doi: 10.1002/jmri.24687.

• Finally, if you want to try and wrap your head around spatial frequency decomposition for image analysis, here is a good overview from the University of New Mexico: https://www.cs.unm.edu/~brayer/vision/fourier.html