Thank you very much Herbie! Just one more question... Since our images have been taken using transmitted light, and the surface of a photographic slide is never absolutely clean (there are always surface deposits/dust, scratches etc), is there a tool that we can use to select only the fungal hyphae of the emulsion surface?
Best Themis Στις Τρί 2 Ιουν 2026 στις 10:48 π.μ., ο/η Gluender <[email protected]> έγραψε: > Greetings Moraitou Themis! > > Without seeing typical images in their original non-lossy file-format > (no JPGs, no screen-shots) it is near to impossible to help. > > "What would be the best approach to improve the separation of hyphae from > the background before thresholding?" > > Frankly speaking: > The best pre-processing is no pre-processing but better images. > > You may try to improve image acquisition by using color lighting for > better contrast. Phase contrast may also be of some help. Avoid > "structures or artifacts are present in the image", etc. > > Best > > Herbie > > ::::::::::::::::::::::::::::::::::::::::::::: > Am 02.06.26 um 11:29 schrieb Moraitou Themis: > > Dear all, > > > > I am interested in quantifying fungal hyphae growth on the surface of > > photographic color slides using images taken with a Dino-Lite microscope. > > > > The basic workflow that I follow in ImageJ roughly includes: > > > > 1. Convert the image to 8-bit grayscale. > > 2. Apply Subtract Background. > > 3. Use automatic thresholding (Otsu and Li seem to give the best results > so > > far). > > > > However, I encounter two main issues: > > > > - The background is not uniform, and some structures or artifacts are > > present in the image. As a result, background subtraction is often not > > sufficient to isolate the hyphae. > > - In some regions, the contrast between the hyphae and the background is > > too low, so the threshold does not detect all hyphae consistently. > > > > What would be the best approach to improve the separation of hyphae from > > the background before thresholding? Are there specific preprocessing > steps, > > filters, color-space transformations, or segmentation methods in > > ImageJ/Fiji that could help achieve more reliable detection and > > quantification? > > > > Any suggestions or examples of similar workflows would be greatly > > appreciated. > > > > Thank you! > > > > -- > > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > > > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html > -- *Efthymia (Themis) Moraitou, PhD* Conservation of Antiquities & Works of Art Cultural Informatics & Communication -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html
