Cleaning tiling artifacts from a z-stack: average intensity projection

 The issue: When scanning an area of tissue with the confocal microscope, many users decide to use the tiling (aka mosaic) option. This can sometimes lead to a visible and sometimes pronounced tiling artifact (aka "football field effect", due to the appearance of visible routed lines in the image) in one or more channels. In other occasions, these artifacts are not easily visible in each individual plane of the z-stack, but it becomes strong when performing a max intensity projection, like in the example in the figures under. The main reason for tiling artifacts is uneven illumination of the field of view. I promise to write a longer post about the reasons later. The main issue of the present post is to find ways to prevent or correct tiling artifacts. There are many strategies to avoid or correct tiling artifacts, and the one mentioned in this post is just one of them.  


Fig. 1: example of a single plane (plane 35) in a 59 plane z-stack, tiled 3x5. Notice that it is already possible to notice a faint tiling artifact at the seems of each tile. 

Fig. 2: Result of a maximum intensity project of the image from fig.1. Notice that the tiling effect becomes strongly enhanced. 

Solution: one possible solution for this problem is something that I found out recently almost accidently. Instead of performing a maximum intensity projection, I created an average intensity projection of the same z-stack, with the surprising effect shown in the figure under. The tiling artifact has practically disappeared. Remember, however, to adjust the visualization histogram of your new image. While maximum intensity projections through the z-plane tends to stretch the pixel values histogram, average intensity projections frequently have the opposite effect, skewing the histogram to  lower levels. Overall, average intensity projections can sometimes reduce background levels and enhance contrast for the relevant biological motif, especially if the z-stack has few levels. However, keep in mind that this solution is not the best in all images, and in some cases it can even lead to watering down important biological features (especially if the z-stack has many levels). 

Fig. 3: Average intensity projection of the image from fig.1, after histogram adjustment. 


 

Kommentarer

Legg inn en kommentar

Populære innlegg fra denne bloggen

Opening Zeiss czi file format mosaic files in Matlab using bio-formats

  The issue: I often write scripts in Matlab to batch image processing tasks, and for that purpose I need to load confocal or axioscan Zeiss czi format image files. I use an open source package called Bio-formats . However, I frequently get two kinds of problems when using this library in Matlab: a) working memory problems (Matlabs Java memory capacity gets filled when the files are in the order of the GB size); and b) stitching problems. This post will only focus on the second issue (I will write later a separate post for the first). The problem is that, for a long time, when I tried to open czi image files of confocal images that were tiled, the bio-formats pack would open the tiles individually, but it would not stitch them as a single image.     Diagnose procedure:  I tried to download different versions of the bio-formats library, because according to internet forums like forum.image.sc , the latest version of bio-formats had that problem fixed. The v...
Les mer

Possible solution for problems opening confocal files from Zeiss LSM 880 in Neurolucida 360 environment

  The issue : One of the regular users of Neurolucida 360 has noticed that some image files produced by the Zeiss LSM 880 (confocal microscope using ZEN black software) do not open. This happens rarely, but she had five examples of medium sized (tens to hundreds of MB sized, mosaics with multiple z-levels) files that NL360 could not open.  Diagnose procedure : I tried to open the same files in Fiji and by using Matlabs bformats pack (v7.0.0) and in both cases, I got error messages concerning corrupted data bytes. Nevertheless it was possible to open those files in ZEN blue lite version. That made me hypothesize that there are a few corrupted data bytes in the file, but that ZEN lite can bypass those data points. That lead me to the think that if I tried to save those files again using the "Save as" menu, the files would be saved with the corrected bytes version, and that would make the files whole again. Upon testing this procedure, the files were fixed and could open normal...
Les mer

Opening multileveled TIFF images in Matlab

  The issue : this is a simple issue. I generally open single-leveled TIFF files in Matlab by simply using the function:    data = imread('example.tif'); However, if the tif file in question is multi-layered (3D), this command will only open the top layer and ignore the rest.  Solution : There are two possible solutions for this problem. 1. Using only one command (tested in MatLab R2023, documentation refers to being first released with R2020b):        data  = tiffreadVolume('example.tif'); The numerical array data will therefore have 3 dimensions (x,y,z). I got it automatically in the format uint8 when I tried. Between the two solutions, this one is the fastest. 2. Using multiple lines (maybe for versions prior to MatLab R2020b):      info = imfinfo('example.tif');   data = zeros(info(1).Width,info(1).Height,numel(info)); %preallocates the data array.   data = uint8(data);  %converts the array from double to uin...
Les mer