One trend that I’ve picked up on in the two years since starting Open MIC is that there’s a huge thirst for learning more about ImageJ/FIJI for processing and analysis. Every time we’ve had a session on ImageJ/FIJI, it’s been a packed room. To meet this demand, Ben Smith from the Vision Sciences Program and I thought it would be great to organize a session on Fourier space and filters. (Ben also wanted to talk about deconvolution, which we didn’t have time to cover.)
Upon first glance, Fourier space seemed like it might be a little too theoretical/abstract for a 90-minute course. However, it’s such a great way to understand resolution and filtering that Ben decided to muscle his way through. He did a great job explaining the theory – and even managed to pull off a humorous slide featuring the Fourier transform equation.
From there, Ben went on to explain how we can use Fourier space to understand how different filters work, specifically focusing on Gaussian, mean, and median filters, as well as the different applications of high pass and low pass filters. Using real world data, he demonstrated how to clean up images with those filters.
Two important take-homes for me were:
- Changing your image data in Fourier space affects the frequency (resolution) but not phase (position). This means you’re applying changes without biasing for positional information — much better in terms of scientific rigor and reproducibility.
- In terms of data management, Ben gave us the tip to name the files with the actual processing steps taken. This way, you know exactly the order in which you applied different filters or thresholds, and so you can repeat them again (if you wish) and have a record of your processing steps.
This was a fantastic session, and we hope to have another one soon on deconvolution. Thanks, Ben!