While attempting to process some of my images from last week, I thought I’d share a cool thing I made (with the help of some others).
This is a three-image composite that was taken in a live rat brain using multiphoton microscopy. The red image is NADH autofluorescence, a metabolite responsible for cellular ATP production. The green image is FAD autofluorescence, which is another metabolite found in cells. The cyan image is (i think) collagen, visualized with second harmonic generation, which I discussed in a previous post here. Not sure what it all means yet, but it sure looks kind of cool.
in vivo images are tricky to work with, especially across multiple types of images with different contrasts. These images proved to be a difficult exercise in ‘image registration’, or trying orienting multiple fields of view into a common coordinate system. This is frequently used by your smartphone to take panoramic images, as well as operating rooms with modern image-guided surgery practices. With microscopes, we’re usually using registration to align repeated images over long time scales, or to align images with multiple tags on them in a single sample. It has a bunch of interesting applications.
The most commonly used alignment methods look for bright and dark spots in the images and look for common positions in multiple images to align them. More advanced methods can look for distance edges and shapes as well as track deformations. But sometimes, you just have to brute force is an do it by hand. That’s sort of what I had to do here…. Nonetheless, it kind of worked.