Hello everyone, a bit of a philosophical discussion on instrumentation here.
Do you find that traditional microscopes have relevance in the age of automated microscopy?
I’m currently working in an industrial preclinical research setting, where fully-automated “confocal imagers” are favored over classical microscopes. They are incredibly efficient for common high-throughput screening applications. However, my background is in traditional confocal microscopy, and I do see where these automated instruments are limiting. They don’t match high-powered microscopes in terms of quality, resolution, and applications requiring flexibility.
For mechanistic projects that require high resolution and active (not automated) sample investigation, a traditional confocal platform suits these needs well.
My work involves investigating rare events. Imagers generally lack eyepieces and interactive motorized stages, so acquisition relies on field sampling. The lack of the ability to progressively scan through magnifications, as one would with a traditional scope, means that these events may be missed-- especially when starting from high magnification, as these imagers do. It seems wasteful to scan an entire sample at several Zs, just to detect rare events, especially when this can be more efficiently achieved with progressive magnification by an active human operator. The same could be applied to histological studies for investigation of specific tissue areas.
Most modern confocal microscope platforms have automatic high-content capabilities, as well, so they offer the best of both worlds.
I realize that this is a broad comparison of two very different imaging approaches, but I’m trying to build a case for the relevance of traditional microscopy, especially in industrial settings.
In my opinion, there is a place for both of these system types in the modern lab, and they can form powerful imaging platforms when used in conjunction.
The difficulty with microscopy is that “inner space” is enormous, as is outer space: Every 10x increase in magnification gives 1000 more places where an image can be captured. Bigger samples are being observed, and super-resolution methods can take you down to a few nanometer resolution. Capturing everything in silico just drowns the observer in data, which still has to be waded through to find the rare events. AI isn’t there yet to do it for you.
Another issue is the complexity of the software. Looking and finding is the hard part of microscopy, but a wall of minute software interface controls can be very intimidating to new users. I manage a core facility which stretches to 9 rooms, and we have had high throughput, high content instruments (Two Pathway HT systems). Their only utility was for screening very large numbers of samples, for chemical genomics and mutant scoring. Most of the science then migrated to “conventional” confocal instruments for human navigation at higher image quality. We had way too much high content capacity for an academic setting, and when they broke down, they were hardly missed. Confocal companies need to put well positions in their base software, to enable multi well plate use; but otherwise a motorized inverted confocal microscope is the main workhorse for microscopy science. Once that is being used flat-out, either get another one, or look at automated boxes to offload the bigger experiments. You may find that camera-based imaging gives better quantitation, and that is more interesting to scientists than the gross speed of operation. So maybe a cheap Keyence is the second need, and a high content box third. It does depend on the science workflow of course.
You have answered your own question by putting your thoughts down in writing - and I agree that it is ‘horses for courses’. The application goal determines the instrument - instruments (should) never lead.
I am a clinical histopathologist - there is no way I could do my job with an ‘imager’ - I need eyes down tubes. I am also a PhD scientist - some of my research could simply not be done without an imaging system - and very specific imaging systems to my research needs.
Some might think: ‘Hold on! Histopathologist? surely ‘digital pathology’ is the ‘thing’ these days for you and that is imager-based’. Unfortunately that is a common misconception these days brought about by ‘hard sell’ marketing by companies and also by academics who depend on ‘digital pathology’ research funds for their livelihoods and career progression prospects - some of whom can be quite persuasive in getting department managers (who are not pathologists / microscopists) to buy into the concept. Sadly the ‘benefits of digital pathology’ are not as clear cut as those vested interest groups make it out to be. I have yet to see a high throughput digital pathology scanner with an image quality coming even close to a correctly set-up Kohler illuminated scope, especially at high power. None of them have easy implementation of things I use routinely on a daily basis (such as polarisation for foreign bodies, fragments of hair shaft, oxalate crystals, etc) - in departments that have ‘converted’ to digital pathology you need to fill out a request to get some birefringence studies - something which I can do in 1 second on my microscope without interruption to my flow of work. I have yet to see an imager system with a simultaneous colour and contrast and gamma range as good as the human eye (yes the pictures look ‘pretty’ but that’s not the point). There are many other reasons why direct vision optical microscopy can never be replaced by digital imagers in particular roles, there is room in this world for both, but for sake of brevity I will stop here.