I want to perform some semi-quantitative analysis of confocal fluorescence images of immunostainning. For example, compare albumin expression on the human hepatocytes and iPS-dervied hepatocytes based on the images of immunostaining. Is there a protocol to make the quantification precise and meaningful? As the fluorescence intensity is not only correlated to albumin expression, but also correlated to confocal plane, excitation light, staining antibody concentration, staining time, and many other factors? And even even fluorescence can be adjusted by several image display mode such as max-min, high constrast..
Is there any good algorithm to calculate fluorescence? For example, some one use DAPI fluorescence as a reference to other fluorescence.
You have already listed a lot of issues with comparing IF intensities... and for those reasons it's recommended to not do it.
If you want to know differences in protein expression, might I suggest doing a Western Blot. I guess you already have an antibody, which might be suitable for WB as well.
Confocal images are good for determining the localization of your protein and for visualization of (sub)cellular structures (like adhesion complexes, microtubules, organelles etc). Such structures can easily be analyzed for size & morphological features in IF-stained samples, but also here, intensities are not reliable.
If you do want to get an idea of a potential difference in expression based on your IF-images (before doing the WB), I would recommend to keep everything similar for your samples:
Seed them in the same glass-plate, fix them for equal time periods, stain with equal incubation times/concentrations, and during imaging use the same settings (laser power, gain, pixel dwell etc). If possible, use a Perfect Focus system to maintain the focal plane, or do a full 3D reconstruction of the cell. When performing measurements, measure in the raw image and do not change contrast/LUTs/intensities etc.
for confocal systems, 3d stacks is essential. This causes some complexity, but it can be handled. for intensity measurements, widefield fluo is better if cover the sample thickness (so 40x-63x for cellular samples is better than 100x objectives).
You won't be able to obtain absolute values, but to compare for example different treatments, it is possible to use confocal. It is hovewer crucial to always use the same instrumentation and procedure (staining protocol, coverglass type, mounting medium etc.), same data acquisition setup (laser line, intesity, detector, Z-stack, resolution... simply everything. I would even recommend to let the laser preheat for the same amount of time - better safe than sorry.) If you can, use highly sensitive detecotrs (Leica's HyDs for example) rather than conventional PMTs, as these produce almost no background noise. I would also recommend not to perform any post-processing on the images and do the quantification on raw data only.
Regarding the quantification, it is better to use a software (for example Cellprofiler) that automatically detects your objects (cells, nuclei etc.) and performs desired quantification, because the level of error would be much more consistent compared to manual dtection.
- Badges
- Science method