The Resolution Series
This post is part of a three-event series on PSF and resolution. Next week another article will be available “How to measure lateral and axial resolution following ISO 21073 and early QUAREP-LIMI recommendations with Argolight Daybook 3”. Two weeks after, the series will close with a webinar with a live demonstration of Daybook new PSF feature.
In preparation of the new Point Spread Function (PSF) feature of Daybook 3 early March 2021, we interviewed core imaging center microscopists about their experience with PSF.
What value do they see in PSF? Why do they do it? How do they do PSF? How often do they do PSF? As you will see in the testimonies the reasons, frequency and application widely vary.
Damien Schapman, is an Engineer and microscopist on the Plate-Forme de Recherche en Imagerie Cellulaire de Normandie (PRIMACEN) in the University of Rouen, FRANCE.
I use PSF for two applications:
- To assess the integrity of an objective, at a given moment and in relation to its previous use.
- To get a second, even third, insight to assess the alignment and pathway of the laser source.
I never do just a PSF. It is always in complement to another test or diagnostic. I use PSF to discriminate the source of an issue. For example, if I have a homogeneity issue or a decrease of illumination power on my sample. Visualizing PSF (in both planes XY and XZ) can confirm or exclude an issue in the path/alignment of the source. On a confocal microscope, I will also run PSF to check on pinhole issues.
When I do PSF, I look at the general shape of the PSF, the tilt, the symmetry, and the numerical values (the full width at half maximum of a Gaussian fit on the PSF). Resolution measurement through the FWHM is important, but I do not only rely on it to assess the state of the microscope/objective. I can have a good FWHM, but a shape that is indicative of bad imaging quality, for example, a banana shape.
I admit that, when everything is fine, when other diagnostics do not raise concerns, I do not take the time to make PSFs. Mainly because it is time-consuming. I need to do about 10 PSFs to get relevant value because, in those ten, I will have outliers. There is always a few too good, a few too bad, truth is in between. So, I average the one in the middle to get a relevant mean value. Outliers come from different reasons: beads get crushed when mounted, irregular drying of the mounting medium creates irregularities in the shape of beads, etc.…
Today I use MetroloJ to analyze my PSF.
Note from Argolight: MetroloJ is a free plugin for ImageJ authored by Fabrice P. Cordelières, Bordeaux Imaging Center (France) and Cédric Matthews, CNRS – IBDML – UMR 6216, Marseille (France). This plugin has been created as part of the metrology workgroup work, within the framework of the “French technological network for multidimensional fluorescence microscopies” (RT- MFM), supported by the Mission ressources et compétences technologiques du CNRS” (MRCT)“.
Link for more information and download: MetroloJ [ImageJ Documentation Wiki] (tudor.lu)
More about the French technological network for multidimensional fluorescence microscopies and its metrology efforts (in French): RTMFM – Microscopie Photonique de Fluorescence Multidimensionnelle (cnrs.fr)
The Dr. Steffen Dietzel is the Head of the Core Facility Bioimaging at the Biomedical Center of the Ludwig-Maximilians-Universität München, in Munich, GERMANY.
We record the PSF:
- To measure experimental resolution.
- To check the performance of microscopes over time.
Since many labs are doing it, it is a good tool to compare results with others.
The PSF gives us an idea of the experimental resolution we can expect, which pixel size to use for Nyquist, and to see how far we are from the theoretical value. Of course, we never reach theoretical values. I recall being alarmed when I did my first PSF, then was reassured when I compared my results with others (and understood the limits of PSF measurements: beads are not points, mounting protocols, etc…). That is why we published some PSF results on our website so that the community and our users have an idea of realistic experimental resolution values.
Ideally, we would perform PSF and other measurements regularly every week or every other week on every system with every objective. Like with a computer data back-up frequency, the question is: how many weeks of work can you afford to lose if a problem occurred? However, we do not have the manpower to do it that often, we actually do it every three months. To avoid this discrepancy between the ideal and the real world, we would need a fully automated process for maintenance checks with test slides implemented in the microscope control software.
What we look for is the size of the PSF (FWHM on Gaussian fit), the tilt, symmetry, and general shape. Rather than the values themselves, it is the stability over time that is important to us.
We developed and published a tool called PSFtracker to evaluate the point spread function of a microscopic system over time. It uses results from the open-source tool PSFJ to generate a display of FWHMs (xy and z) over time and to create images of typical beads/PSFs in xy, xz and yz (More info and example of tracking here). This visual representation allows us to easily detect banana-shaped PSFs, stage drift, and the like. If a fluctuation is detected, we perform additional tests to figure out what the problem is (objective damage, laser alignment, other). If we cannot solve it, we call technical service.
I would not call the PSF a catch-all indicator – homogeneity is a good example of a thing not shown by PSF – but rather a catch-many: if the PSF fluctuates it is a good indicator that something is wrong and further investigation is warranted.
We once picked up a major increase of the FWHM with PSF tracker, but no user noticed the issue, nobody complained about imaging quality. We sent an e-mail around: check your data because they may be wrong. If we would have done PSFs only every six months, we would have had to come back at every image made in the past six months. It is not a good feeling if people record bad images at your facility, so we do these regular maintenance checks to ensure a stable high-quality imaging environment.
Dr. Aurélien Dauphin is a Biologist, Microscopist, Bioimage analyst at the Cell and Tissue Imaging Facility in Paris, FRANCE.
I do PSFs regularly. However, the frequency of my PSFs measurement varies according to the use of the objectives and microscopes. Most of my measurements are made on high magnification objectives (40x and higher) every 4-6 months.
I do them for three reasons:
- To know the actual resolution.
- To know the spherical aberrations and objectives flaws.
- To get a PSF to input into deconvolution algorithms.
The frequency at which I do PSF is related to their usage: in the facility, we have an intense usage of live imaging microscopes, involving long time-lapses on inverted microscopes with oil-immersion or water-immersion objectives. After some time, the oil might end up entering the objective, causing imaging issues.
This is when the PSF is useful: it gives a broad view of the whole quality of the microscope, including some oil/cell medium spilled around. We have to clean the objective holder, or even deeper in the microscope, and sometimes we end up sending objectives for repair.
The oil issue is frequent depending on the brand (different objective front lens rim designs) and if there is an oil recovery system. But eventually, it’s either the microscope or the objective that gets filled with oil (either behind the front lens or even deeper lenses).
I do at least 5 to 10 beads per objective (usually 200-100 nm beads) because there is a disparity in the results. I look for the size of the PSF (FWHM on Gaussian fit) and the general shape (symmetry and tilt).
The frequency is quite low compared to other quality control because we experienced that the PSF fluctuated less than field illumination or laser power. In comparison, we check laser power every 2-6 months (according to the microscope).
Finally, for in vivo study, we also do multichromatic beads (100 nm Tetraspeck beads) because we have a multi-camera, multi-channel system. But in this case, we use the beads for channel alignment, so we do not really extract PSF measurements out of this, but rather chromatic and camera position shift.
I use MetroloJ for my analysis.
Daybook 3 Troubleshooting new PSF feature will be available for free: easily and quickly measure FWHM on PSF Z-stacks.
Export values as CSV or ready to print reports. Upgrade for Daybook 3 Quality Control to get PSF value tracking, table, graphs, and value thresholds.
The Resolution Series This post is part of a three-event series on PSF and resolution. First part is titled "Why do you do PSF : 3 Experts...
The Argo-POWERHM offers many ways to monitor the illumination power of fluorescence microscopes. With the integration in Daybook Analysis, live...
How to use the Point Spread Function (PSF) analysis in Daybook (Webinar) During this webinar, Gautier PAPON from Argolight, will present the...