Bright field imageexplained

Maybe a last comment about the data, you seem to have some horizontal bands on the images too… this may be linked to your having neon or LED lights on in the room while you acquire, and you are seeing the 50 Hz frequency of the ambien light…

Do you think the image background could somehow be improved so that the contrast could be increased without introducing artefacts? (I cannot repeat the assay so these images are all I have to work with )

Bright field imageexample

Unfortunately I would have to agree with that remark, and unfortunately I am afraid that no amount of preprocessing is going to be able to help you. These are issues that should be adressed on the image acquisition and microscopy side.

If you’re doing this visually, you can blur the image a lot using a Gaussian Filter and dividing the original image by it.

Bright fieldmicroscope diagram

Another issue that the images are JPEG compressed and the JPEG artifacts are roughly the size of your cells. Those artifacts will be enhanced as well… Ideally the camera attached to the microscope can save images in an uncompressed way. This is something your group should look into if they wish to do automated quantitative analysis using that microscope.

The background itself is not a problem, it’s the lack of unique features to distinguish cells from artifacts that is lacking. Most methods would rely on enhancing edges, but considering you have about as many artifacts as you have cells, they will be equally enhanced and you’ve moved the problem down to “how to reduce the other artifacts now that the contrast is better”. It’s a Russian Dolls kind of problem…

Bright field imageinterpretation

Bright fieldmicroscope

I am sorry about not being able to provide you with better news, especially considering that you have 95 images to go through, but other than manually annotating the areas, I have nothing else to offer for this data.

In the case that a user would come to our facility with such data, we would be forced to refuse helping them analyze it, and would proceed to help them optimize their imaging, usually with a small batch and some quick testing before ramping up the acquisition to a full experiment.

Darkfieldmicroscope

In Fiji, press CTRL + SHIFT + N (Or CMD instead of CTRL on Mac) to bring up the script editor, and copy paste the code below, then hit run. It will ask you for a file, and you should point it to your image.

Hello. I am trying to preprocess some images for segmentation using FIJI but I am struggling due to noise, especially light artefacts. The images were obtained using a brightfield microscope. The purpose of the images is to investigate cell migration, so I am interested in the gap between the two sides rather than quantifying cell counts. I have 95 images and have outlined most of them manually, but the contrast is so poor and there is so much noise that I highly doubt the accuracy of my approach, so I would like to know if the images can be preprocessed such that I could accurately used an automated segmentation method. I have carried out a pseudo flat field correction using the BioVoxxel plugin to address the shading discrepancy in some of the images, but I am not sure what else to do. Thank you in advance for any help!