I am using a high-res build. With the achromatic lens ( Achromatic lens for OpenFlexure Microscopes – IO Rodeo ) and 40x objective ( 40X Plan Objective Microscope Lens – IO Rodeo ), the captured background illumination shows a distinct, repetitive shading pattern of vertical dark “finger-like” bands or ribs along the four edges of the frame (more on the left and right). Any ideas of where the obstruction comes from?
strong text
The image is also very blue. I have a few questions:
Is this after camera calibration? If it is, could you disable the flat field and lens shading (where that is in the webapp depends on the software version) and then post the uncorrected image here.
A photo of the microscope overall can also help to diagnose some problems.
Which version of the software are you using?
Those screenshots are from the camera stage mapping step, which comes after the camera calibration.
Those images do show however that the illumination seems to be off-centre, which will be a problem.
I think that it also shows software v2, which is the current stable release. We have just moved the new software v3 to Beta status, there is a thread on the forum. The calibration sequence is greatly improved. The calibration wizard shows what you need to do more clearly, and the calibration algorithms are better as well. If the Pi that you are using is a Pi4, and has at least 4GB RAM, I think it would be useful to re-flash your SD card with the v3.0.0-beta1 software.
Excellent.
The background now looks very uniform, and you have a sample that is in focus. The camera stage mapping is very likely to fail with that sample, it has a strong edge between blue and clear, but there is very little other detail. You want to find a slide with a high contrast details over an area of several fields of view for both x and y.
Prepared plant stem specimens can be good, or a permanent marker scribbled onto a blank slide.
Will get better samples for testing. I appreciate your help.
The example image is a sample that appears to be prepared differently, but most importantly will have used an objective with higher magnification and higher NA. Red blood cells are small, and to image them well you need very high resolution, and the resolution limit is set by the NA.
The 40× lens that you reference in your first post is a fairly low cost, reasonable quality lens with NA0.65. For a really clear image of blood cells one would usually use an oil immersion lens with NA 1.25, probably 100×. The lens is also only part of the requirement, the illumination also needs to be well focussed and well aligned to be able to illuminate at high NA.
The out of focus edges are either alignment, or the lens performance. The ‘plan’ specification of your lens designates that it has been designed to try to achieve focus across all of the image at the same time, but this is difficult to design and lower cost ‘plan’ lenses still can have different focus at the edges. If you adjust the focus slightly, do you see the centre in focus changing to a ring around the edge being in focus? If so, that is the performance of the lens. The other effect could be tilt of the optics module. That would show up if when you adjust the focus one edge is in focus, then a strip across the middle, then the other edge.
For example Focusing and Chromatic Aberration Issues - #7 by zengzpi uses a high quality lens (probably costing over $1000) and careful alignment for excellent resolution. Other more subtle image quality issues are discussed in that thread.
Thank you, @WilliamW for your detailed reply.
I changed the objective to Nikon 50X/0.80.
The image quality has slightly improved but is still not satisfactory.
I have not got the blood microscopy experience to really tell what to expect from that lens and sample.
In general, at 0.80NA the illumination focus and centering become important. That can take trial and error to sort out the best alignment. It is also worth checking the diffuser in the illumination. A damaged or degraded diffuser can adversely affect imaging, even at lower NA than this (Illumination and resolution issues (#447) · Issues · OpenFlexure / openflexure-microscope · GitLab, Illumination/Diffuser problems (#448) · Issues · OpenFlexure / openflexure-microscope · GitLab).
I think you have been through the obvious things on the imaging side - the objective and tube lens. The optics module body is different for finite or infinite conjugate objectives. The marking that we can see on the Nikon lens do not show whether it is designed for finite conjugates.
Thank you so much @WilliamW! Every reply you have provided has been incredibly important, and I really appreciate the knowledge you’re sharing.
Your mention of finite vs. infinite systems made me double-check my setup, and I think I might have made a serious mistake.
Right now, I am using a plan finite objective (40X Plan Objective Microscope Lens – IO Rodeo), but I also added this 50mm achromatic lens (Achromatic lens for OpenFlexure Microscopes – IO Rodeo) into the optical path.
Am I correct in assuming that because I am using a finite objective, I should completely remove this achromatic lens from my setup? My understanding now is that finite objectives project light directly to the sensor without needing a tube lens, and leaving it in will ruin the image quality.
You do need the 50mm tube lens.
The finite corrected objectives are designed to focus quite a long way away, and to produce an image that is 17-20mm across. We need to demagnify that image to fit on the small Pi camera sensor, which is what the 50mm tube lens does. This is detailed in build.openflexure.org/openflexure-microscope/v7.0.0-beta5/info_pages/imaging_optics_explanation.html.
The position of the tube lens is slightly different for finite or infinite conjugates objectives, so there are two slightly different versions of the optics modules. I think you got yours in a kit, in which case it will match the finite conjugates objective that you got at the same time.
The only things that could be wrong there woild be if the tube lens is not pushed home fully in the gripper, or if it is in the wrong way up. They will only make small differences, but small differences is what you are looking for.
Interestingly, I got really good images of bird red blood cells using a Zeiss infinite oil lens while keeping the achromatic tube lens in place (supposedly optimized for finite lenses).
As always, the color is less saturated at the edges compared with the center. Was it due to the light source (a bad condenser or diffuser) or the camera (Pi 2 camera)?
Other than that, the image looks great. The bird cells are about 7-10 micrometers in short diameter.
This is because of the construction of the camera, to work with the wide angle lens that it is supplied with. The effect is particularly noticeable in red/purple samples.
The issue, characterisation and solutions are detailed in Flat-Field and Colour Correction for the Raspberry Pi Camera Module | Journal of Open Hardware It is possible to post-process images to correct for the colour mixing, but not live.
The Pi HQ camera does not have this problem. In the current OpenFlexure Server v3.0.0-beta1 it is possible to make some manual configuration changes to use the HQ camera. It is on the software roadmap to make this a more integrated configuration experience.