We are currently measuring mitochondrial length, width, area, type (vesicular or non-vesicular) within electron microscopy images. Since this is performed by manually drawing around each mitochondrion in a given image, it is terribly time consuming. We are looking for an automated method. Unfortunately these mitochondria are not labelled in any way. Since the EM images are inherently noisy, we are not sure how to effectively detect mitochondria. I tried with ImageJ but it didn't work. Any suggestions?
I tried many stuff (even programming it myself), nothing works with enough reliability to substitute the manual profile drawing, is not easy to train anything to distinguish vacoules, autophagosomes from mitochondria... For helping you in drawing profiles recommend you the TrakEM plugin, it also allows you to review what you marked and to store it so you have a good record of data, if at some point you need to go back to it. https://imagej.net/TrakEM2
I know it's tedious but once you get some experience you can be pretty fast, a touchscreen or a drawing pad helps a lot; I was at about 2-3 animals/day in this article https://journals.plos.org/plosone/article/figure?id=10.1371/journal.pone.0018568.g007
cheers!
- Badges
- Science method