Measuring small sizes
Locked
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Remko, Thank you for the prompt response to the PSF thread. I have another question that is somewhat related to the topic. We need to measure diameters of small intracellular bubbles that can be less than a micron across. Ours is a regular (not superresolution) system - either widefield fluorescence, or confocal fluorescence, or (hopefully, in the near future) a Fluoview 3000 with deconvolution. Has anyone proved that accurate measurements on such a small scale are possible? What would it take? Some kind of deconvolution - 2D? 3D? Does anyone know what others have done in this direction? Thank you Mike Model |
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** I don't know if this is a common technique, but I like the idea of rather than solving the ill-posed inversion problem, attacking the well-posed forward problem: IF this is my system PSF, and IF this is my belief about my object's diameter, then what measurement do I expect? How does that compare to what I actually measured? If you want to formalize it, you could express it as a maximum likelihood problem with Poisson noise. On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Remko, > > Thank you for the prompt response to the PSF thread. > > I have another question that is somewhat related to the topic. We need to > measure diameters of small intracellular bubbles that can be less than a > micron across. Ours is a regular (not superresolution) system - either > widefield fluorescence, or confocal fluorescence, or (hopefully, in the > near future) a Fluoview 3000 with deconvolution. Has anyone proved that > accurate measurements on such a small scale are possible? What would it > take? Some kind of deconvolution - 2D? 3D? Does anyone know what others > have done in this direction? Thank you > > Mike Model > |
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** I see. Only you would probably need a good criterion of what constitutes the best match. -----Original Message----- From: Confocal Microscopy List <[hidden email]> On Behalf Of Andrew York Sent: Friday, August 17, 2018 3:44 PM To: [hidden email] Subject: Re: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** I don't know if this is a common technique, but I like the idea of rather than solving the ill-posed inversion problem, attacking the well-posed forward problem: IF this is my system PSF, and IF this is my belief about my object's diameter, then what measurement do I expect? How does that compare to what I actually measured? If you want to formalize it, you could express it as a maximum likelihood problem with Poisson noise. On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Remko, > > Thank you for the prompt response to the PSF thread. > > I have another question that is somewhat related to the topic. We need > to measure diameters of small intracellular bubbles that can be less > than a micron across. Ours is a regular (not superresolution) system - > either widefield fluorescence, or confocal fluorescence, or > (hopefully, in the near future) a Fluoview 3000 with deconvolution. > Has anyone proved that accurate measurements on such a small scale are > possible? What would it take? Some kind of deconvolution - 2D? 3D? > Does anyone know what others have done in this direction? Thank you > > Mike Model > |
|
|
In reply to this post by Andrew York
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Well this might be useful? Kong et al Sub-microscopic analysis of t-tubule geometry in living cardiac ventricular myocytes using a shape-based analysis method. J Mol Cell Cardiol. 2017;108:1–7. HTH Mark B. Cannell. Ph.D. FRSNZ FISHR Department of Physiology, Pharmacology & Neuroscience School of Medical Sciences University Walk Bristol BS8 1TD [hidden email] On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew York" <[hidden email] on behalf of [hidden email]> wrote: ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** I don't know if this is a common technique, but I like the idea of rather than solving the ill-posed inversion problem, attacking the well-posed forward problem: IF this is my system PSF, and IF this is my belief about my object's diameter, then what measurement do I expect? How does that compare to what I actually measured? If you want to formalize it, you could express it as a maximum likelihood problem with Poisson noise. On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Remko, > > Thank you for the prompt response to the PSF thread. > > I have another question that is somewhat related to the topic. We need to > measure diameters of small intracellular bubbles that can be less than a > micron across. Ours is a regular (not superresolution) system - either > widefield fluorescence, or confocal fluorescence, or (hopefully, in the > near future) a Fluoview 3000 with deconvolution. Has anyone proved that > accurate measurements on such a small scale are possible? What would it > take? Some kind of deconvolution - 2D? 3D? Does anyone know what others > have done in this direction? Thank you > > Mike Model > |
|
|
In reply to this post by mmodel
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** In a 2015 paper (https://doi.org/10.1371/journal.pone.0134112) we reported on the localization of beads ranging from 50 nm to 1 um in diameter. The localization was carried out by a maximum likelihood fitting of an Airy PSF to the image of the bead. The data model used included both Poisson shot noise and Gaussian measurement (readout) noise. As part of the fitting, the width of the Airy PSF was also estimated in some cases. While the width of the Airy PSF is not the bead diameter, it should provide an indication of what the diameter is. In the same paper a model was also presented that describes the image of a bead as a sphere convolved with a 3D PSF. Even though this model was used with the bead diameter as a known quantity in our maximum likelihood fitting, it can in principle also be used to estimate the diameter itself. Hope this at least provides some ideas for estimating the diameter of very small objects. Best regards, Jerry Jerry Chao, Ph.D. Department of Biomedical Engineering Texas A&M University Email: [hidden email] |
 
|
Cromey, Douglas W - (dcromey) |
Re: Measuring small sizes
|
|
In reply to this post by mmodel
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Michael, Are these bubbles a different RI than the cystoplasm? Several years ago I worked with a group looking at microbubbles and they had a tendency to bend the light (spherical object of different refractive index) in the images, making sizing difficult. Doug ------------------------------------------------------------------------------------------ Douglas W. Cromey, M.S. - Associate Scientific Investigator Dept. of Cellular & Molecular Medicine, University of Arizona 1501 N. Campbell Ave, Tucson, AZ 85724-5044 USA office: LSN 463 email: [hidden email] voice: 520-626-2824 fax: 520-626-2097 http://microscopy.arizona.edu/learn/microscopy-imaging-resources-www Home of: "Microscopy and Imaging Resources on the WWW" UA Microscopy Alliance - http://microscopy.arizona.edu -----Original Message----- From: Confocal Microscopy List <[hidden email]> On Behalf Of MODEL, MICHAEL Sent: Friday, August 17, 2018 11:31 AM To: [hidden email] Subject: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Remko, Thank you for the prompt response to the PSF thread. I have another question that is somewhat related to the topic. We need to measure diameters of small intracellular bubbles that can be less than a micron across. Ours is a regular (not superresolution) system - either widefield fluorescence, or confocal fluorescence, or (hopefully, in the near future) a Fluoview 3000 with deconvolution. Has anyone proved that accurate measurements on such a small scale are possible? What would it take? Some kind of deconvolution - 2D? 3D? Does anyone know what others have done in this direction? Thank you Mike Model |
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Yes, and that's the reason we want to know their sizes - to measure refractive indices (= protein concentration) ________________________________ From: Confocal Microscopy List <[hidden email]> on behalf of Cromey, Douglas W - (dcromey) <[hidden email]> Sent: Friday, August 17, 2018 6:52 PM To: [hidden email] Subject: Re: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU<http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> lists.umn.edu [hidden email]: listserv archives. confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Michael, Are these bubbles a different RI than the cystoplasm? Several years ago I worked with a group looking at microbubbles and they had a tendency to bend the light (spherical object of different refractive index) in the images, making sizing difficult. Doug ------------------------------------------------------------------------------------------ Douglas W. Cromey, M.S. - Associate Scientific Investigator Dept. of Cellular & Molecular Medicine, University of Arizona 1501 N. Campbell Ave, Tucson, AZ 85724-5044 USA office: LSN 463 email: [hidden email] voice: 520-626-2824 fax: 520-626-2097 http://microscopy.arizona.edu/learn/microscopy-imaging-resources-www Home of: "Microscopy and Imaging Resources on the WWW" UA Microscopy Alliance - http://microscopy.arizona.edu -----Original Message----- From: Confocal Microscopy List <[hidden email]> On Behalf Of MODEL, MICHAEL Sent: Friday, August 17, 2018 11:31 AM To: [hidden email] Subject: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Remko, Thank you for the prompt response to the PSF thread. I have another question that is somewhat related to the topic. We need to measure diameters of small intracellular bubbles that can be less than a micron across. Ours is a regular (not superresolution) system - either widefield fluorescence, or confocal fluorescence, or (hopefully, in the near future) a Fluoview 3000 with deconvolution. Has anyone proved that accurate measurements on such a small scale are possible? What would it take? Some kind of deconvolution - 2D? 3D? Does anyone know what others have done in this direction? Thank you Mike Model |
|
|
In reply to this post by Jerry Chao
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Jerry, Is your code written in Matlab? ________________________________ From: Confocal Microscopy List <[hidden email]> on behalf of Jerry Chao <[hidden email]> Sent: Friday, August 17, 2018 6:30 PM To: [hidden email] Subject: Re: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU<http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> lists.umn.edu [hidden email]: listserv archives. confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** In a 2015 paper (https://doi.org/10.1371/journal.pone.0134112) we reported on the localization of beads ranging from 50 nm to 1 um in diameter. The localization was carried out by a maximum likelihood fitting of an Airy PSF to the image of the bead. The data model used included both Poisson shot noise and Gaussian measurement (readout) noise. As part of the fitting, the width of the Airy PSF was also estimated in some cases. While the width of the Airy PSF is not the bead diameter, it should provide an indication of what the diameter is. In the same paper a model was also presented that describes the image of a bead as a sphere convolved with a 3D PSF. Even though this model was used with the bead diameter as a known quantity in our maximum likelihood fitting, it can in principle also be used to estimate the diameter itself. Hope this at least provides some ideas for estimating the diameter of very small objects. Best regards, Jerry Jerry Chao, Ph.D. Department of Biomedical Engineering Texas A&M University Email: [hidden email] |
|
|
In reply to this post by Mark Cannell-2
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Yes, thank you ________________________________ From: Confocal Microscopy List <[hidden email]> on behalf of Mark Cannell <[hidden email]> Sent: Friday, August 17, 2018 4:23 PM To: [hidden email] Subject: Re: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU<http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> lists.umn.edu [hidden email]: listserv archives. confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Well this might be useful? Kong et al Sub-microscopic analysis of t-tubule geometry in living cardiac ventricular myocytes using a shape-based analysis method. J Mol Cell Cardiol. 2017;108:1–7. HTH Mark B. Cannell. Ph.D. FRSNZ FISHR Department of Physiology, Pharmacology & Neuroscience School of Medical Sciences University Walk Bristol BS8 1TD [hidden email] On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew York" <[hidden email] on behalf of [hidden email]> wrote: ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** I don't know if this is a common technique, but I like the idea of rather than solving the ill-posed inversion problem, attacking the well-posed forward problem: IF this is my system PSF, and IF this is my belief about my object's diameter, then what measurement do I expect? How does that compare to what I actually measured? If you want to formalize it, you could express it as a maximum likelihood problem with Poisson noise. On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Remko, > > Thank you for the prompt response to the PSF thread. > > I have another question that is somewhat related to the topic. We need to > measure diameters of small intracellular bubbles that can be less than a > micron across. Ours is a regular (not superresolution) system - either > widefield fluorescence, or confocal fluorescence, or (hopefully, in the > near future) a Fluoview 3000 with deconvolution. Has anyone proved that > accurate measurements on such a small scale are possible? What would it > take? Some kind of deconvolution - 2D? 3D? Does anyone know what others > have done in this direction? Thank you > > Mike Model > |
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Hi Michael, On the topic of modelling your object to extract structural parameters, we have done some work along exactly those lines. From super-resolution images to study spherical HSV1 viruses: https://www.nature.com/articles/ncomms6980 But the method can be equally applied to wide-field (non-SR) images. It goes along the lines of what Andrew was suggesting. Some more work on model-fitting could also be helpful to you, it's ELM, and generalises the approach that we used on the viruses: https://www.sciencedirect.com/science/article/pii/S0006349515009935 I hope this helps. Best, Romain ____________________________________ *Dr Romain F. Laine*, PhD in Biophotonics PDRA at the Quantitative Imaging and Nanobiophysics group MRC Laboratory of Molecular Cell Biology (LMCB) University College London Gower Street, London, WC1E 6BT http://www.ucl.ac.uk/lmcb/users/romain-laine On Sat, Aug 18, 2018 at 12:56 PM MODEL, MICHAEL <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Yes, thank you > > > ________________________________ > From: Confocal Microscopy List <[hidden email]> on > behalf of Mark Cannell <[hidden email]> > Sent: Friday, August 17, 2018 4:23 PM > To: [hidden email] > Subject: Re: Measuring small sizes > > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU< > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> > lists.umn.edu > [hidden email]: listserv archives. confocalmicroscopy > > > > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Well this might be useful? > > Kong et al Sub-microscopic analysis of t-tubule geometry in living > cardiac ventricular myocytes using a shape-based analysis method. J Mol > Cell Cardiol. 2017;108:1–7. > > HTH > > Mark B. Cannell. Ph.D. FRSNZ FISHR > Department of Physiology, Pharmacology & Neuroscience > School of Medical Sciences > University Walk > Bristol BS8 1TD > > [hidden email] > > > > On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew York" > <[hidden email] on behalf of > [hidden email]> wrote: > > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your > posting. > ***** > > I don't know if this is a common technique, but I like the idea of > rather > than solving the ill-posed inversion problem, attacking the well-posed > forward problem: > > IF this is my system PSF, and IF this is my belief about my object's > diameter, then what measurement do I expect? How does that compare to > what > I actually measured? > > If you want to formalize it, you could express it as a maximum > likelihood > problem with Poisson noise. > > On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL <[hidden email]> > wrote: > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Remko, > > > > Thank you for the prompt response to the PSF thread. > > > > I have another question that is somewhat related to the topic. We > need to > > measure diameters of small intracellular bubbles that can be less > than a > > micron across. Ours is a regular (not superresolution) system - > either > > widefield fluorescence, or confocal fluorescence, or (hopefully, in > the > > near future) a Fluoview 3000 with deconvolution. Has anyone proved > that > > accurate measurements on such a small scale are possible? What would > it > > take? Some kind of deconvolution - 2D? 3D? Does anyone know what > others > > have done in this direction? Thank you > > > > Mike Model > > > > > |
|
|
In reply to this post by mmodel
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Hello Michael, Yes, our code is written in Matlab. The code for the maximum likelihood fitting of the Airy PSF is part of our EstimationTool software package. EstimationTool is freely available, and you can obtain a copy through the following URL if you'd like to try it. http://wardoberlab.com/software/estimationtool The maximum likelihood fitting will allow you to estimate the width of the Airy PSF using a Poisson noise model or a Poisson and Gaussian noise model. The 2D Gaussian PSF is supported as well, but with either PSF the estimated width will somehow need to be translated to the diameter you're looking for. Best regards, Jerry |
|
|
In reply to this post by Romain Laine
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Hi Mike Deconvolution is useful for reducing blur caused by the PSF and improving contrast and subsequent performance of segmentation and in general improves measurements, but you have to be careful interpreting very small spatial measurements after deconvolution. As suggested by others, a fitting scheme is a good approach. One issue is that the Full Width at Half Max of small objects after deconvolution, will be different depending on the parameters used in the algorithm. If I deconvolve with classic Richardson Lucy, the FWHM after decon get's smaller with increasing iterations. If I use regularization the FWHM after decon is different depending on the regularization factor. Claire Brown has examined measurements of 2.5um diameter microspheres after deconvolution. https://www.researchgate.net/publication/261257717_Calibrati on_of_Wide-Field_Deconvolution_Microscopy_for_Quantitative_ Fluorescence_Imaging Her paper mostly focused on examining whether intensities were still linear after deconvolution, however she also took volumetric measurements and recorded the effect of both fixed theoretical and blind PSF on these measurements. Claire measured the volume, not diameter of the spheres, but found that the volume varied quite a bit depending on the PSF used ( expected volume 8 um^3 , Huygens fixed PSF deconvolution gave a deconvolved volume of ~ 10 um^3, Autoquant fixed 12 um^3, and Autoquant Blind ~ 13 um^3, volume measured without deconvolution was 18 um^3) An accurate PSF is important if you want the best measurement, if you can't measure the PSF, as you noted in your paper on Spherical Aberration and Deconvolution ( https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2818.2010.03416.x) a theoretical PSF with educated guess for sample RI and distance from coverslip works well. Other factors such as the optimization approach and parameters (ie number of iterations and regularization factors) may change the apparent size of objects after deconvolution. It would be interesting to see similar experiments using measured PSFs. What would also be interesting is to do a similar experiment but vary the size of the bead instead of intensity. Maybe start with 2.5um beads and then use increasingly smaller beads and look at the relationship between true bead size, measurement from the original image and measurement from the deconvolved image. Remko has Huygen's done such experiments?? Brian On Sat, Aug 18, 2018 at 6:15 AM, Romain Laine <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Hi Michael, > On the topic of modelling your object to extract structural parameters, we > have done some work along exactly those lines. > From super-resolution images to study spherical HSV1 viruses: > https://www.nature.com/articles/ncomms6980 > > But the method can be equally applied to wide-field (non-SR) images. It > goes along the lines of what Andrew was suggesting. > > Some more work on model-fitting could also be helpful to you, it's ELM, and > generalises the approach that we used on the viruses: > https://www.sciencedirect.com/science/article/pii/S0006349515009935 > > I hope this helps. > > Best, > > Romain > > > ____________________________________ > > *Dr Romain F. Laine*, PhD in Biophotonics > > PDRA at the Quantitative Imaging and Nanobiophysics group > > MRC Laboratory of Molecular Cell Biology (LMCB) > > University College London > > Gower Street, London, WC1E 6BT > > http://www.ucl.ac.uk/lmcb/users/romain-laine > > On Sat, Aug 18, 2018 at 12:56 PM MODEL, MICHAEL <[hidden email]> wrote: > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Yes, thank you > > > > > > ________________________________ > > From: Confocal Microscopy List <[hidden email]> on > > behalf of Mark Cannell <[hidden email]> > > Sent: Friday, August 17, 2018 4:23 PM > > To: [hidden email] > > Subject: Re: Measuring small sizes > > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > > > LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU< > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> > > lists.umn.edu > > [hidden email]: listserv archives. confocalmicroscopy > > > > > > > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Well this might be useful? > > > > Kong et al Sub-microscopic analysis of t-tubule geometry in living > > cardiac ventricular myocytes using a shape-based analysis method. J Mol > > Cell Cardiol. 2017;108:1–7. > > > > HTH > > > > Mark B. Cannell. Ph.D. FRSNZ FISHR > > Department of Physiology, Pharmacology & Neuroscience > > School of Medical Sciences > > University Walk > > Bristol BS8 1TD > > > > [hidden email] > > > > > > > > On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew > York" > > <[hidden email] on behalf of > > [hidden email]> wrote: > > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > > posting. > > ***** > > > > I don't know if this is a common technique, but I like the idea of > > rather > > than solving the ill-posed inversion problem, attacking the > well-posed > > forward problem: > > > > IF this is my system PSF, and IF this is my belief about my object's > > diameter, then what measurement do I expect? How does that compare to > > what > > I actually measured? > > > > If you want to formalize it, you could express it as a maximum > > likelihood > > problem with Poisson noise. > > > > On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL <[hidden email]> > > wrote: > > > > > ***** > > > To join, leave or search the confocal microscopy listserv, go to: > > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > > Post images on http://www.imgur.com and include the link in your > > posting. > > > ***** > > > > > > Remko, > > > > > > Thank you for the prompt response to the PSF thread. > > > > > > I have another question that is somewhat related to the topic. We > > need to > > > measure diameters of small intracellular bubbles that can be less > > than a > > > micron across. Ours is a regular (not superresolution) system - > > either > > > widefield fluorescence, or confocal fluorescence, or (hopefully, in > > the > > > near future) a Fluoview 3000 with deconvolution. Has anyone proved > > that > > > accurate measurements on such a small scale are possible? What > would > > it > > > take? Some kind of deconvolution - 2D? 3D? Does anyone know what > > others > > > have done in this direction? Thank you > > > > > > Mike Model > > > > > > > > > > |
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Brian, Thank you for replying and for the important reference. Perhaps if the PSF is more or less cylindrical, without too much mushrooming above and below focus, then a 2D deconvolution (with an assumption of spherical shape) would be adequate? Mike -----Original Message----- From: Confocal Microscopy List <[hidden email]> On Behalf Of Brian Northan Sent: Tuesday, August 21, 2018 9:03 AM To: [hidden email] Subject: Re: Measuring small sizes ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Hi Mike Deconvolution is useful for reducing blur caused by the PSF and improving contrast and subsequent performance of segmentation and in general improves measurements, but you have to be careful interpreting very small spatial measurements after deconvolution. As suggested by others, a fitting scheme is a good approach. One issue is that the Full Width at Half Max of small objects after deconvolution, will be different depending on the parameters used in the algorithm. If I deconvolve with classic Richardson Lucy, the FWHM after decon get's smaller with increasing iterations. If I use regularization the FWHM after decon is different depending on the regularization factor. Claire Brown has examined measurements of 2.5um diameter microspheres after deconvolution. https://www.researchgate.net/publication/261257717_Calibrati on_of_Wide-Field_Deconvolution_Microscopy_for_Quantitative_ Fluorescence_Imaging Her paper mostly focused on examining whether intensities were still linear after deconvolution, however she also took volumetric measurements and recorded the effect of both fixed theoretical and blind PSF on these measurements. Claire measured the volume, not diameter of the spheres, but found that the volume varied quite a bit depending on the PSF used ( expected volume 8 um^3 , Huygens fixed PSF deconvolution gave a deconvolved volume of ~ 10 um^3, Autoquant fixed 12 um^3, and Autoquant Blind ~ 13 um^3, volume measured without deconvolution was 18 um^3) An accurate PSF is important if you want the best measurement, if you can't measure the PSF, as you noted in your paper on Spherical Aberration and Deconvolution ( https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2818.2010.03416.x) a theoretical PSF with educated guess for sample RI and distance from coverslip works well. Other factors such as the optimization approach and parameters (ie number of iterations and regularization factors) may change the apparent size of objects after deconvolution. It would be interesting to see similar experiments using measured PSFs. What would also be interesting is to do a similar experiment but vary the size of the bead instead of intensity. Maybe start with 2.5um beads and then use increasingly smaller beads and look at the relationship between true bead size, measurement from the original image and measurement from the deconvolved image. Remko has Huygen's done such experiments?? Brian On Sat, Aug 18, 2018 at 6:15 AM, Romain Laine <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Hi Michael, > On the topic of modelling your object to extract structural > parameters, we have done some work along exactly those lines. > From super-resolution images to study spherical HSV1 viruses: > https://www.nature.com/articles/ncomms6980 > > But the method can be equally applied to wide-field (non-SR) images. > It goes along the lines of what Andrew was suggesting. > > Some more work on model-fitting could also be helpful to you, it's > ELM, and generalises the approach that we used on the viruses: > https://www.sciencedirect.com/science/article/pii/S0006349515009935 > > I hope this helps. > > Best, > > Romain > > > ____________________________________ > > *Dr Romain F. Laine*, PhD in Biophotonics > > PDRA at the Quantitative Imaging and Nanobiophysics group > > MRC Laboratory of Molecular Cell Biology (LMCB) > > University College London > > Gower Street, London, WC1E 6BT > > http://www.ucl.ac.uk/lmcb/users/romain-laine > > On Sat, Aug 18, 2018 at 12:56 PM MODEL, MICHAEL <[hidden email]> wrote: > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Yes, thank you > > > > > > ________________________________ > > From: Confocal Microscopy List <[hidden email]> on > > behalf of Mark Cannell <[hidden email]> > > Sent: Friday, August 17, 2018 4:23 PM > > To: [hidden email] > > Subject: Re: Measuring small sizes > > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > > > LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU< > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> > > lists.umn.edu > > [hidden email]: listserv archives. > > confocalmicroscopy > > > > > > > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Well this might be useful? > > > > Kong et al Sub-microscopic analysis of t-tubule geometry in living > > cardiac ventricular myocytes using a shape-based analysis method. J > > Mol Cell Cardiol. 2017;108:1–7. > > > > HTH > > > > Mark B. Cannell. Ph.D. FRSNZ FISHR > > Department of Physiology, Pharmacology & Neuroscience School of > > Medical Sciences University Walk Bristol BS8 1TD > > > > [hidden email] > > > > > > > > On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew > York" > > <[hidden email] on behalf of > > [hidden email]> wrote: > > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > > posting. > > ***** > > > > I don't know if this is a common technique, but I like the idea > > of rather > > than solving the ill-posed inversion problem, attacking the > well-posed > > forward problem: > > > > IF this is my system PSF, and IF this is my belief about my object's > > diameter, then what measurement do I expect? How does that > > compare to what > > I actually measured? > > > > If you want to formalize it, you could express it as a maximum > > likelihood > > problem with Poisson noise. > > > > On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL > > <[hidden email]> > > wrote: > > > > > ***** > > > To join, leave or search the confocal microscopy listserv, go to: > > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > > Post images on http://www.imgur.com and include the link in > > your posting. > > > ***** > > > > > > Remko, > > > > > > Thank you for the prompt response to the PSF thread. > > > > > > I have another question that is somewhat related to the topic. > > We need to > > > measure diameters of small intracellular bubbles that can be > > less than a > > > micron across. Ours is a regular (not superresolution) system > > - either > > > widefield fluorescence, or confocal fluorescence, or > > (hopefully, in the > > > near future) a Fluoview 3000 with deconvolution. Has anyone > > proved that > > > accurate measurements on such a small scale are possible? What > would > > it > > > take? Some kind of deconvolution - 2D? 3D? Does anyone know > > what others > > > have done in this direction? Thank you > > > > > > Mike Model > > > > > > > > > > |
|
|
Remko Dijkstra |
Re: Measuring small sizes **Vendor reply**
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** ** Vendor reply ** Hi Mike, This is an interesting thread, and various useful approaches have already been suggested. A maximum likelihood estimation (MLE) --or similar-- approach is overall recommended. The limitation of using a 2D 'fitting' approach, is that any object that is (slightly) out-of-focus might not be fitted correctly. Unless your objects are always located in the same plane, this could potentially bias your fitting results. Thus, algorithms with 3D fitting capabilities, would definitely add value over any 2D fitting method when determining object sizes. As Brian also suggested: you could also consider deconvolution: the main deconvolution algorithms in Huygens are all advanced 3D MLE based: Quick MLE, Classic MLE, Good's Roughness MLE. These algorithms take Poisson noise into account, and use a regularization parameter that is directly related to the SNR in the image. The estimate found by the MLE algorithms must be regarded as a best estimation of the object. So a good 3D MLE deconvolution approach has many of the suggested properties mentioned in this discussion. Additional advantage: the MLE deconvolution approach is well accepted in the microscopy community, considering the long list of papers that have used MLE-based deconvolution algorithms to process microscopy images over the past two decades. The approach that Andrew initially described is in fact quite similar to 3D MLE deconvolution, but with the additional 'a priori' knowledge input of the size/shape of your objects. In turn this is very similar to what our PSF distiller does: as 'a-priori' knowledge of the size of the beads is used as input, along with the SNR of the measured bead image. The PSF distiller then uses a 3D MLE approach to find the most likely PSF that formed the measured image, given the size of the (spherical) objects and image SNR. Simply put, this process (which we call PSF distilling) is the opposite of what you do with MLE deconvolution, where the algorithm finds the most likely intensity distribution (object) that formed the measured image with the given PSF and given SNR. Since the (MLE) deconvolved result is already a maximum-likelihood model of your object(s): measurements on the object size, shape and position will be much more reliable compared to measuring on raw data alone. With Huygens Professional it is also possible to simulate images with a given PSF + Poisson Noise. Simulated images can be generated using your specific microscope's measured PSF, and images can be created with objects of various (known) sizes. These simulations can then be used as a control/calibration for your actual measurements by deconvolving the simulated images with the exact same settings as your measured images. Since the object sizes in your simulated images are known, you could compare and generate a 'look-up table' to match the measured sizes after deconvolution with the actual (known) object sizes from the simulation. Of course there will always be a (practical) limit as to how far you can go with small size measurements and conventional light microscopy. The obvious limit is of course directly related to the diffraction limit. Using MLE deconvolution you will be able to restore and measure sizes below the conventional diffraction limit, but it will still be limited nevertheless. For some additional practical considerations I would like to refer to: https://svi.nl/ObjectSize Lastly, if possible I would recommend to minimize RI mismatch between the lens and embedding medium: probably by using a water or glycerol objective. This way you could also measure the PSF with beads near the coverslip, distill the PSF, and then use the distilled PSF for deconvolution. This way you can minimize any RI-mismatch effects, and improve the reliability of the size measurements of small particles. And as Doug already pointed out: if the RI of the bubbles themselves are significantly different than the surrounding RI, and if they are large enough, then they could act like small lenses within your sample. This will obviously make matters much more complex! Related to Brian's suggestion and question: next to the simulations that I mentioned, you could indeed also image beads of different sizes for comparison/calibration, but it would be important to use beads that have similar RI as the embedding medium as to prevent any lens effects. In the previous 'Tetraspeck beads for PSFs' thread on this list, I referred to a paper on methods for confocal axial calibration by Besseling et.al. : https://doi.org/10.1111/jmi.12194 In this same paper they also present size measurement results (including deconvolution) on PMMA beads of different sizes. Kind regards, Remko *********************************************************** Remko Dijkstra, MSc Imaging Specialist/Account Manager Scientific Volume Imaging bv Tel: + 31 35 642 1626 www.svi.nl *********************************************************** For support matters contact: [hidden email] Op 8/21/2018 om 4:10 PM schreef MODEL, MICHAEL: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Brian, > Thank you for replying and for the important reference. Perhaps if the PSF is more or less cylindrical, without too much mushrooming above and below focus, then a 2D deconvolution (with an assumption of spherical shape) would be adequate? > > Mike > > -----Original Message----- > From: Confocal Microscopy List <[hidden email]> On Behalf Of Brian Northan > Sent: Tuesday, August 21, 2018 9:03 AM > To: [hidden email] > Subject: Re: Measuring small sizes > > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Hi Mike > > Deconvolution is useful for reducing blur caused by the PSF and improving contrast and subsequent performance of segmentation and in general improves measurements, but you have to be careful interpreting very small spatial measurements after deconvolution. As suggested by others, a fitting scheme is a good approach. > > One issue is that the Full Width at Half Max of small objects after deconvolution, will be different depending on the parameters used in the algorithm. If I deconvolve with classic Richardson Lucy, the FWHM after decon get's smaller with increasing iterations. If I use regularization the FWHM after decon is different depending on the regularization factor. > > Claire Brown has examined measurements of 2.5um diameter microspheres after deconvolution. > > https://www.researchgate.net/publication/261257717_Calibrati > on_of_Wide-Field_Deconvolution_Microscopy_for_Quantitative_ > Fluorescence_Imaging > > Her paper mostly focused on examining whether intensities were still linear after deconvolution, however she also took volumetric measurements and recorded the effect of both fixed theoretical and blind PSF on these measurements. > > Claire measured the volume, not diameter of the spheres, but found that the volume varied quite a bit depending on the PSF used ( expected volume 8 um^3 , Huygens fixed PSF deconvolution gave a deconvolved volume of ~ 10 um^3, Autoquant fixed 12 um^3, and Autoquant Blind ~ 13 um^3, volume measured without deconvolution was 18 um^3) > > An accurate PSF is important if you want the best measurement, if you can't measure the PSF, as you noted in your paper on Spherical Aberration and Deconvolution ( > https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2818.2010.03416.x) > a theoretical PSF with educated guess for sample RI and distance from coverslip works well. > > Other factors such as the optimization approach and parameters (ie number of iterations and regularization factors) may change the apparent size of objects after deconvolution. > > It would be interesting to see similar experiments using measured PSFs. > > What would also be interesting is to do a similar experiment but vary the size of the bead instead of intensity. Maybe start with 2.5um beads and then use increasingly smaller beads and look at the relationship between true bead size, measurement from the original image and measurement from the deconvolved image. > > Remko has Huygen's done such experiments?? > > Brian > > On Sat, Aug 18, 2018 at 6:15 AM, Romain Laine <[hidden email]> wrote: > >> ***** >> To join, leave or search the confocal microscopy listserv, go to: >> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >> Post images on http://www.imgur.com and include the link in your posting. >> ***** >> >> Hi Michael, >> On the topic of modelling your object to extract structural >> parameters, we have done some work along exactly those lines. >> From super-resolution images to study spherical HSV1 viruses: >> https://www.nature.com/articles/ncomms6980 >> >> But the method can be equally applied to wide-field (non-SR) images. >> It goes along the lines of what Andrew was suggesting. >> >> Some more work on model-fitting could also be helpful to you, it's >> ELM, and generalises the approach that we used on the viruses: >> https://www.sciencedirect.com/science/article/pii/S0006349515009935 >> >> I hope this helps. >> >> Best, >> >> Romain >> >> >> ____________________________________ >> >> *Dr Romain F. Laine*, PhD in Biophotonics >> >> PDRA at the Quantitative Imaging and Nanobiophysics group >> >> MRC Laboratory of Molecular Cell Biology (LMCB) >> >> University College London >> >> Gower Street, London, WC1E 6BT >> >> http://www.ucl.ac.uk/lmcb/users/romain-laine >> >> On Sat, Aug 18, 2018 at 12:56 PM MODEL, MICHAEL <[hidden email]> wrote: >> >>> ***** >>> To join, leave or search the confocal microscopy listserv, go to: >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> Post images on http://www.imgur.com and include the link in your >> posting. >>> ***** >>> >>> Yes, thank you >>> >>> >>> ________________________________ >>> From: Confocal Microscopy List <[hidden email]> on >>> behalf of Mark Cannell <[hidden email]> >>> Sent: Friday, August 17, 2018 4:23 PM >>> To: [hidden email] >>> Subject: Re: Measuring small sizes >>> >>> ***** >>> To join, leave or search the confocal microscopy listserv, go to: >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> >>> LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU< >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> >>> lists.umn.edu >>> [hidden email]: listserv archives. >>> confocalmicroscopy >>> >>> >>> >>> Post images on http://www.imgur.com and include the link in your >> posting. >>> ***** >>> >>> Well this might be useful? >>> >>> Kong et al Sub-microscopic analysis of t-tubule geometry in living >>> cardiac ventricular myocytes using a shape-based analysis method. J >>> Mol Cell Cardiol. 2017;108:1–7. >>> >>> HTH >>> >>> Mark B. Cannell. Ph.D. FRSNZ FISHR >>> Department of Physiology, Pharmacology & Neuroscience School of >>> Medical Sciences University Walk Bristol BS8 1TD >>> >>> [hidden email] >>> >>> >>> >>> On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew >> York" >>> <[hidden email] on behalf of >>> [hidden email]> wrote: >>> >>> ***** >>> To join, leave or search the confocal microscopy listserv, go to: >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> Post images on http://www.imgur.com and include the link in your >>> posting. >>> ***** >>> >>> I don't know if this is a common technique, but I like the idea >>> of rather >>> than solving the ill-posed inversion problem, attacking the >> well-posed >>> forward problem: >>> >>> IF this is my system PSF, and IF this is my belief about my object's >>> diameter, then what measurement do I expect? How does that >>> compare to what >>> I actually measured? >>> >>> If you want to formalize it, you could express it as a maximum >>> likelihood >>> problem with Poisson noise. >>> >>> On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL >>> <[hidden email]> >>> wrote: >>> >>> > ***** >>> > To join, leave or search the confocal microscopy listserv, go to: >>> > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> > Post images on http://www.imgur.com and include the link in >>> your posting. >>> > ***** >>> > >>> > Remko, >>> > >>> > Thank you for the prompt response to the PSF thread. >>> > >>> > I have another question that is somewhat related to the topic. >>> We need to >>> > measure diameters of small intracellular bubbles that can be >>> less than a >>> > micron across. Ours is a regular (not superresolution) system >>> - either >>> > widefield fluorescence, or confocal fluorescence, or >>> (hopefully, in the >>> > near future) a Fluoview 3000 with deconvolution. Has anyone >>> proved that >>> > accurate measurements on such a small scale are possible? What >> would >>> it >>> > take? Some kind of deconvolution - 2D? 3D? Does anyone know >>> what others >>> > have done in this direction? Thank you >>> > >>> > Mike Model >>> > >>> >>> >>> |
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Remko, I certainly understand that the entire 3d object cannot be correctly restored through a 2d deconvolution, but it is less obvious to me that its lateral diameter also cannot be restored. (If the brightest portion of the PSF if cylindrical enough within the thickness of the object). Am I wrong? Also, even though refractive bubbles certainly act as lenses for external light (the fact we are going to use quite extensively in the other part of the project), this is probably less important for fluorescence coming from inside the bubbles because the rays cross their boundaries more or less radially. In any event, there is nothing one can do about it because the vesicles with higher refractive index are embedded in the cytosol and one cannot match both. Mike -----Original Message----- From: Confocal Microscopy List <[hidden email]> On Behalf Of Remko Dijkstra Sent: Wednesday, August 22, 2018 10:57 AM To: [hidden email] Subject: Re: Measuring small sizes **Vendor reply** ***** To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** ** Vendor reply ** Hi Mike, This is an interesting thread, and various useful approaches have already been suggested. A maximum likelihood estimation (MLE) --or similar-- approach is overall recommended. The limitation of using a 2D 'fitting' approach, is that any object that is (slightly) out-of-focus might not be fitted correctly. Unless your objects are always located in the same plane, this could potentially bias your fitting results. Thus, algorithms with 3D fitting capabilities, would definitely add value over any 2D fitting method when determining object sizes. As Brian also suggested: you could also consider deconvolution: the main deconvolution algorithms in Huygens are all advanced 3D MLE based: Quick MLE, Classic MLE, Good's Roughness MLE. These algorithms take Poisson noise into account, and use a regularization parameter that is directly related to the SNR in the image. The estimate found by the MLE algorithms must be regarded as a best estimation of the object. So a good 3D MLE deconvolution approach has many of the suggested properties mentioned in this discussion. Additional advantage: the MLE deconvolution approach is well accepted in the microscopy community, considering the long list of papers that have used MLE-based deconvolution algorithms to process microscopy images over the past two decades. The approach that Andrew initially described is in fact quite similar to 3D MLE deconvolution, but with the additional 'a priori' knowledge input of the size/shape of your objects. In turn this is very similar to what our PSF distiller does: as 'a-priori' knowledge of the size of the beads is used as input, along with the SNR of the measured bead image. The PSF distiller then uses a 3D MLE approach to find the most likely PSF that formed the measured image, given the size of the (spherical) objects and image SNR. Simply put, this process (which we call PSF distilling) is the opposite of what you do with MLE deconvolution, where the algorithm finds the most likely intensity distribution (object) that formed the measured image with the given PSF and given SNR. Since the (MLE) deconvolved result is already a maximum-likelihood model of your object(s): measurements on the object size, shape and position will be much more reliable compared to measuring on raw data alone. With Huygens Professional it is also possible to simulate images with a given PSF + Poisson Noise. Simulated images can be generated using your specific microscope's measured PSF, and images can be created with objects of various (known) sizes. These simulations can then be used as a control/calibration for your actual measurements by deconvolving the simulated images with the exact same settings as your measured images. Since the object sizes in your simulated images are known, you could compare and generate a 'look-up table' to match the measured sizes after deconvolution with the actual (known) object sizes from the simulation. Of course there will always be a (practical) limit as to how far you can go with small size measurements and conventional light microscopy. The obvious limit is of course directly related to the diffraction limit. Using MLE deconvolution you will be able to restore and measure sizes below the conventional diffraction limit, but it will still be limited nevertheless. For some additional practical considerations I would like to refer to: https://svi.nl/ObjectSize Lastly, if possible I would recommend to minimize RI mismatch between the lens and embedding medium: probably by using a water or glycerol objective. This way you could also measure the PSF with beads near the coverslip, distill the PSF, and then use the distilled PSF for deconvolution. This way you can minimize any RI-mismatch effects, and improve the reliability of the size measurements of small particles. And as Doug already pointed out: if the RI of the bubbles themselves are significantly different than the surrounding RI, and if they are large enough, then they could act like small lenses within your sample. This will obviously make matters much more complex! Related to Brian's suggestion and question: next to the simulations that I mentioned, you could indeed also image beads of different sizes for comparison/calibration, but it would be important to use beads that have similar RI as the embedding medium as to prevent any lens effects. In the previous 'Tetraspeck beads for PSFs' thread on this list, I referred to a paper on methods for confocal axial calibration by Besseling et.al. : https://doi.org/10.1111/jmi.12194 In this same paper they also present size measurement results (including deconvolution) on PMMA beads of different sizes. Kind regards, Remko *********************************************************** Remko Dijkstra, MSc Imaging Specialist/Account Manager Scientific Volume Imaging bv Tel: + 31 35 642 1626 www.svi.nl *********************************************************** For support matters contact: [hidden email] Op 8/21/2018 om 4:10 PM schreef MODEL, MICHAEL: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Brian, > Thank you for replying and for the important reference. Perhaps if the PSF is more or less cylindrical, without too much mushrooming above and below focus, then a 2D deconvolution (with an assumption of spherical shape) would be adequate? > > Mike > > -----Original Message----- > From: Confocal Microscopy List <[hidden email]> On > Behalf Of Brian Northan > Sent: Tuesday, August 21, 2018 9:03 AM > To: [hidden email] > Subject: Re: Measuring small sizes > > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Hi Mike > > Deconvolution is useful for reducing blur caused by the PSF and improving contrast and subsequent performance of segmentation and in general improves measurements, but you have to be careful interpreting very small spatial measurements after deconvolution. As suggested by others, a fitting scheme is a good approach. > > One issue is that the Full Width at Half Max of small objects after deconvolution, will be different depending on the parameters used in the algorithm. If I deconvolve with classic Richardson Lucy, the FWHM after decon get's smaller with increasing iterations. If I use regularization the FWHM after decon is different depending on the regularization factor. > > Claire Brown has examined measurements of 2.5um diameter microspheres after deconvolution. > > https://www.researchgate.net/publication/261257717_Calibrati > on_of_Wide-Field_Deconvolution_Microscopy_for_Quantitative_ > Fluorescence_Imaging > > Her paper mostly focused on examining whether intensities were still linear after deconvolution, however she also took volumetric measurements and recorded the effect of both fixed theoretical and blind PSF on these measurements. > > Claire measured the volume, not diameter of the spheres, but found > that the volume varied quite a bit depending on the PSF used ( > expected volume 8 um^3 , Huygens fixed PSF deconvolution gave a > deconvolved volume of ~ 10 um^3, Autoquant fixed 12 um^3, and > Autoquant Blind ~ 13 um^3, volume measured without deconvolution was > 18 um^3) > > An accurate PSF is important if you want the best measurement, if you > can't measure the PSF, as you noted in your paper on Spherical > Aberration and Deconvolution ( > https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2818.2010.0341 > 6.x) a theoretical PSF with educated guess for sample RI and distance > from coverslip works well. > > Other factors such as the optimization approach and parameters (ie number of iterations and regularization factors) may change the apparent size of objects after deconvolution. > > It would be interesting to see similar experiments using measured PSFs. > > What would also be interesting is to do a similar experiment but vary the size of the bead instead of intensity. Maybe start with 2.5um beads and then use increasingly smaller beads and look at the relationship between true bead size, measurement from the original image and measurement from the deconvolved image. > > Remko has Huygen's done such experiments?? > > Brian > > On Sat, Aug 18, 2018 at 6:15 AM, Romain Laine <[hidden email]> wrote: > >> ***** >> To join, leave or search the confocal microscopy listserv, go to: >> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >> Post images on http://www.imgur.com and include the link in your posting. >> ***** >> >> Hi Michael, >> On the topic of modelling your object to extract structural >> parameters, we have done some work along exactly those lines. >> From super-resolution images to study spherical HSV1 viruses: >> https://www.nature.com/articles/ncomms6980 >> >> But the method can be equally applied to wide-field (non-SR) images. >> It goes along the lines of what Andrew was suggesting. >> >> Some more work on model-fitting could also be helpful to you, it's >> ELM, and generalises the approach that we used on the viruses: >> https://www.sciencedirect.com/science/article/pii/S0006349515009935 >> >> I hope this helps. >> >> Best, >> >> Romain >> >> >> ____________________________________ >> >> *Dr Romain F. Laine*, PhD in Biophotonics >> >> PDRA at the Quantitative Imaging and Nanobiophysics group >> >> MRC Laboratory of Molecular Cell Biology (LMCB) >> >> University College London >> >> Gower Street, London, WC1E 6BT >> >> http://www.ucl.ac.uk/lmcb/users/romain-laine >> >> On Sat, Aug 18, 2018 at 12:56 PM MODEL, MICHAEL <[hidden email]> wrote: >> >>> ***** >>> To join, leave or search the confocal microscopy listserv, go to: >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> Post images on http://www.imgur.com and include the link in your >> posting. >>> ***** >>> >>> Yes, thank you >>> >>> >>> ________________________________ >>> From: Confocal Microscopy List <[hidden email]> on >>> behalf of Mark Cannell <[hidden email]> >>> Sent: Friday, August 17, 2018 4:23 PM >>> To: [hidden email] >>> Subject: Re: Measuring small sizes >>> >>> ***** >>> To join, leave or search the confocal microscopy listserv, go to: >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> >>> LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU< >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> >>> lists.umn.edu >>> [hidden email]: listserv archives. >>> confocalmicroscopy >>> >>> >>> >>> Post images on http://www.imgur.com and include the link in your >> posting. >>> ***** >>> >>> Well this might be useful? >>> >>> Kong et al Sub-microscopic analysis of t-tubule geometry in living >>> cardiac ventricular myocytes using a shape-based analysis method. J >>> Mol Cell Cardiol. 2017;108:1–7. >>> >>> HTH >>> >>> Mark B. Cannell. Ph.D. FRSNZ FISHR >>> Department of Physiology, Pharmacology & Neuroscience School of >>> Medical Sciences University Walk Bristol BS8 1TD >>> >>> [hidden email] >>> >>> >>> >>> On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew >> York" >>> <[hidden email] on behalf of >>> [hidden email]> wrote: >>> >>> ***** >>> To join, leave or search the confocal microscopy listserv, go to: >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> Post images on http://www.imgur.com and include the link in >>> your posting. >>> ***** >>> >>> I don't know if this is a common technique, but I like the idea >>> of rather >>> than solving the ill-posed inversion problem, attacking the >> well-posed >>> forward problem: >>> >>> IF this is my system PSF, and IF this is my belief about my object's >>> diameter, then what measurement do I expect? How does that >>> compare to what >>> I actually measured? >>> >>> If you want to formalize it, you could express it as a maximum >>> likelihood >>> problem with Poisson noise. >>> >>> On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL >>> <[hidden email]> >>> wrote: >>> >>> > ***** >>> > To join, leave or search the confocal microscopy listserv, go to: >>> > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy >>> > Post images on http://www.imgur.com and include the link in >>> your posting. >>> > ***** >>> > >>> > Remko, >>> > >>> > Thank you for the prompt response to the PSF thread. >>> > >>> > I have another question that is somewhat related to the topic. >>> We need to >>> > measure diameters of small intracellular bubbles that can be >>> less than a >>> > micron across. Ours is a regular (not superresolution) system >>> - either >>> > widefield fluorescence, or confocal fluorescence, or >>> (hopefully, in the >>> > near future) a Fluoview 3000 with deconvolution. Has anyone >>> proved that >>> > accurate measurements on such a small scale are possible? >>> What >> would >>> it >>> > take? Some kind of deconvolution - 2D? 3D? Does anyone know >>> what others >>> > have done in this direction? Thank you >>> > >>> > Mike Model >>> > >>> >>> >>> |
|
|
Brian Northan |
Re: Measuring small sizes **Vendor reply**
|
|
*****
To join, leave or search the confocal microscopy listserv, go to: http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy Post images on http://www.imgur.com and include the link in your posting. ***** Hi Mike The issue with a 2D approach as Remko mentioned (deconvolution or fitting) is simply the fact that some objects could be slightly out of focus, and that would effect the measurement. Brian On Wed, Aug 22, 2018 at 10:06 AM MODEL, MICHAEL <[hidden email]> wrote: > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > Remko, > > I certainly understand that the entire 3d object cannot be correctly > restored through a 2d deconvolution, but it is less obvious to me that its > lateral diameter also cannot be restored. (If the brightest portion of the > PSF if cylindrical enough within the thickness of the object). Am I wrong? > > Also, even though refractive bubbles certainly act as lenses for external > light (the fact we are going to use quite extensively in the other part of > the project), this is probably less important for fluorescence coming from > inside the bubbles because the rays cross their boundaries more or less > radially. In any event, there is nothing one can do about it because the > vesicles with higher refractive index are embedded in the cytosol and one > cannot match both. > > Mike > > -----Original Message----- > From: Confocal Microscopy List <[hidden email]> On > Behalf Of Remko Dijkstra > Sent: Wednesday, August 22, 2018 10:57 AM > To: [hidden email] > Subject: Re: Measuring small sizes **Vendor reply** > > ***** > To join, leave or search the confocal microscopy listserv, go to: > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > Post images on http://www.imgur.com and include the link in your posting. > ***** > > ** Vendor reply ** > > Hi Mike, > > This is an interesting thread, and various useful approaches have already > been suggested. A maximum likelihood estimation (MLE) --or > similar-- approach is overall recommended. > The limitation of using a 2D 'fitting' approach, is that any object that > is (slightly) out-of-focus might not be fitted correctly. Unless your > objects are always located in the same plane, this could potentially bias > your fitting results. > Thus, algorithms with 3D fitting capabilities, would definitely add value > over any 2D fitting method when determining object sizes. > > As Brian also suggested: you could also consider deconvolution: the main > deconvolution algorithms in Huygens are all advanced 3D MLE based: Quick > MLE, Classic MLE, Good's Roughness MLE. These algorithms take Poisson noise > into account, and use a regularization parameter that is directly related > to the SNR in the image. > The estimate found by the MLE algorithms must be regarded as a best > estimation of the object. So a good 3D MLE deconvolution approach has many > of the suggested properties mentioned in this discussion. > Additional advantage: the MLE deconvolution approach is well accepted in > the microscopy community, considering the long list of papers that have > used MLE-based deconvolution algorithms to process microscopy images over > the past two decades. > > The approach that Andrew initially described is in fact quite similar to > 3D MLE deconvolution, but with the additional 'a priori' knowledge input of > the size/shape of your objects. > In turn this is very similar to what our PSF distiller does: as 'a-priori' > knowledge of the size of the beads is used as input, along with the SNR of > the measured bead image. The PSF distiller then uses a 3D MLE approach to > find the most likely PSF that formed the measured image, given the size of > the (spherical) objects and image SNR. Simply put, this process (which we > call PSF distilling) is the opposite of what you do with MLE deconvolution, > where the algorithm finds the most likely intensity distribution (object) > that formed the measured image with the given PSF and given SNR. > > Since the (MLE) deconvolved result is already a maximum-likelihood model > of your object(s): measurements on the object size, shape and position will > be much more reliable compared to measuring on raw data alone. > > With Huygens Professional it is also possible to simulate images with a > given PSF + Poisson Noise. Simulated images can be generated using your > specific microscope's measured PSF, and images can be created with objects > of various (known) sizes. These simulations can then be used as a > control/calibration for your actual measurements by deconvolving the > simulated images with the exact same settings as your measured images. > Since the object sizes in your simulated images are known, you could > compare and generate a 'look-up table' to match the measured sizes after > deconvolution with the actual (known) object sizes from the simulation. > > Of course there will always be a (practical) limit as to how far you can > go with small size measurements and conventional light microscopy. The > obvious limit is of course directly related to the diffraction limit. > Using MLE deconvolution you will be able to restore and measure sizes > below the conventional diffraction limit, but it will still be limited > nevertheless. For some additional practical considerations I would like to > refer to: https://svi.nl/ObjectSize > > Lastly, if possible I would recommend to minimize RI mismatch between the > lens and embedding medium: probably by using a water or glycerol objective. > This way you could also measure the PSF with beads near the coverslip, > distill the PSF, and then use the distilled PSF for deconvolution. > This way you can minimize any RI-mismatch effects, and improve the > reliability of the size measurements of small particles. And as Doug > already pointed out: if the RI of the bubbles themselves are significantly > different than the surrounding RI, and if they are large enough, then they > could act like small lenses within your sample. This will obviously make > matters much more complex! > Related to Brian's suggestion and question: next to the simulations that I > mentioned, you could indeed also image beads of different sizes for > comparison/calibration, but it would be important to use beads that have > similar RI as the embedding medium as to prevent any lens effects. > In the previous 'Tetraspeck beads for PSFs' thread on this list, I > referred to a paper on methods for confocal axial calibration by Besseling > et.al. : https://doi.org/10.1111/jmi.12194 In this same paper they also > present size measurement results (including > deconvolution) on PMMA beads of different sizes. > > Kind regards, > > Remko > > *********************************************************** > Remko Dijkstra, MSc > Imaging Specialist/Account Manager > Scientific Volume Imaging bv > Tel: + 31 35 642 1626 > www.svi.nl > *********************************************************** > For support matters contact: [hidden email] > > Op 8/21/2018 om 4:10 PM schreef MODEL, MICHAEL: > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Brian, > > Thank you for replying and for the important reference. Perhaps if the > PSF is more or less cylindrical, without too much mushrooming above and > below focus, then a 2D deconvolution (with an assumption of spherical > shape) would be adequate? > > > > Mike > > > > -----Original Message----- > > From: Confocal Microscopy List <[hidden email]> On > > Behalf Of Brian Northan > > Sent: Tuesday, August 21, 2018 9:03 AM > > To: [hidden email] > > Subject: Re: Measuring small sizes > > > > ***** > > To join, leave or search the confocal microscopy listserv, go to: > > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > > Post images on http://www.imgur.com and include the link in your > posting. > > ***** > > > > Hi Mike > > > > Deconvolution is useful for reducing blur caused by the PSF and > improving contrast and subsequent performance of segmentation and in > general improves measurements, but you have to be careful interpreting very > small spatial measurements after deconvolution. As suggested by others, a > fitting scheme is a good approach. > > > > One issue is that the Full Width at Half Max of small objects after > deconvolution, will be different depending on the parameters used in the > algorithm. If I deconvolve with classic Richardson Lucy, the FWHM after > decon get's smaller with increasing iterations. If I use regularization > the FWHM after decon is different depending on the regularization factor. > > > > Claire Brown has examined measurements of 2.5um diameter microspheres > after deconvolution. > > > > https://www.researchgate.net/publication/261257717_Calibrati > > on_of_Wide-Field_Deconvolution_Microscopy_for_Quantitative_ > > Fluorescence_Imaging > > > > Her paper mostly focused on examining whether intensities were still > linear after deconvolution, however she also took volumetric measurements > and recorded the effect of both fixed theoretical and blind PSF on these > measurements. > > > > Claire measured the volume, not diameter of the spheres, but found > > that the volume varied quite a bit depending on the PSF used ( > > expected volume 8 um^3 , Huygens fixed PSF deconvolution gave a > > deconvolved volume of ~ 10 um^3, Autoquant fixed 12 um^3, and > > Autoquant Blind ~ 13 um^3, volume measured without deconvolution was > > 18 um^3) > > > > An accurate PSF is important if you want the best measurement, if you > > can't measure the PSF, as you noted in your paper on Spherical > > Aberration and Deconvolution ( > > https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2818.2010.0341 > > 6.x) a theoretical PSF with educated guess for sample RI and distance > > from coverslip works well. > > > > Other factors such as the optimization approach and parameters (ie > number of iterations and regularization factors) may change the apparent > size of objects after deconvolution. > > > > It would be interesting to see similar experiments using measured PSFs. > > > > What would also be interesting is to do a similar experiment but vary > the size of the bead instead of intensity. Maybe start with 2.5um beads > and then use increasingly smaller beads and look at the relationship > between true bead size, measurement from the original image and measurement > from the deconvolved image. > > > > Remko has Huygen's done such experiments?? > > > > Brian > > > > On Sat, Aug 18, 2018 at 6:15 AM, Romain Laine <[hidden email]> > wrote: > > > >> ***** > >> To join, leave or search the confocal microscopy listserv, go to: > >> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > >> Post images on http://www.imgur.com and include the link in your > posting. > >> ***** > >> > >> Hi Michael, > >> On the topic of modelling your object to extract structural > >> parameters, we have done some work along exactly those lines. > >> From super-resolution images to study spherical HSV1 viruses: > >> https://www.nature.com/articles/ncomms6980 > >> > >> But the method can be equally applied to wide-field (non-SR) images. > >> It goes along the lines of what Andrew was suggesting. > >> > >> Some more work on model-fitting could also be helpful to you, it's > >> ELM, and generalises the approach that we used on the viruses: > >> https://www.sciencedirect.com/science/article/pii/S0006349515009935 > >> > >> I hope this helps. > >> > >> Best, > >> > >> Romain > >> > >> > >> ____________________________________ > >> > >> *Dr Romain F. Laine*, PhD in Biophotonics > >> > >> PDRA at the Quantitative Imaging and Nanobiophysics group > >> > >> MRC Laboratory of Molecular Cell Biology (LMCB) > >> > >> University College London > >> > >> Gower Street, London, WC1E 6BT > >> > >> http://www.ucl.ac.uk/lmcb/users/romain-laine > >> > >> On Sat, Aug 18, 2018 at 12:56 PM MODEL, MICHAEL <[hidden email]> > wrote: > >> > >>> ***** > >>> To join, leave or search the confocal microscopy listserv, go to: > >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > >>> Post images on http://www.imgur.com and include the link in your > >> posting. > >>> ***** > >>> > >>> Yes, thank you > >>> > >>> > >>> ________________________________ > >>> From: Confocal Microscopy List <[hidden email]> on > >>> behalf of Mark Cannell <[hidden email]> > >>> Sent: Friday, August 17, 2018 4:23 PM > >>> To: [hidden email] > >>> Subject: Re: Measuring small sizes > >>> > >>> ***** > >>> To join, leave or search the confocal microscopy listserv, go to: > >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > >>> > >>> LISTSERV 16.0 - CONFOCALMICROSCOPY List at LISTS.UMN.EDU< > >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> > >>> lists.umn.edu > >>> [hidden email]: listserv archives. > >>> confocalmicroscopy > >>> > >>> > >>> > >>> Post images on http://www.imgur.com and include the link in your > >> posting. > >>> ***** > >>> > >>> Well this might be useful? > >>> > >>> Kong et al Sub-microscopic analysis of t-tubule geometry in living > >>> cardiac ventricular myocytes using a shape-based analysis method. J > >>> Mol Cell Cardiol. 2017;108:1–7. > >>> > >>> HTH > >>> > >>> Mark B. Cannell. Ph.D. FRSNZ FISHR > >>> Department of Physiology, Pharmacology & Neuroscience School of > >>> Medical Sciences University Walk Bristol BS8 1TD > >>> > >>> [hidden email] > >>> > >>> > >>> > >>> On 17/08/18, 8:44 PM, "Confocal Microscopy List on behalf of Andrew > >> York" > >>> <[hidden email] on behalf of > >>> [hidden email]> wrote: > >>> > >>> ***** > >>> To join, leave or search the confocal microscopy listserv, go to: > >>> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > >>> Post images on http://www.imgur.com and include the link in > >>> your posting. > >>> ***** > >>> > >>> I don't know if this is a common technique, but I like the idea > >>> of rather > >>> than solving the ill-posed inversion problem, attacking the > >> well-posed > >>> forward problem: > >>> > >>> IF this is my system PSF, and IF this is my belief about my > object's > >>> diameter, then what measurement do I expect? How does that > >>> compare to what > >>> I actually measured? > >>> > >>> If you want to formalize it, you could express it as a maximum > >>> likelihood > >>> problem with Poisson noise. > >>> > >>> On Fri, Aug 17, 2018 at 11:31 AM, MODEL, MICHAEL > >>> <[hidden email]> > >>> wrote: > >>> > >>> > ***** > >>> > To join, leave or search the confocal microscopy listserv, go > to: > >>> > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy > >>> > Post images on http://www.imgur.com and include the link in > >>> your posting. > >>> > ***** > >>> > > >>> > Remko, > >>> > > >>> > Thank you for the prompt response to the PSF thread. > >>> > > >>> > I have another question that is somewhat related to the topic. > >>> We need to > >>> > measure diameters of small intracellular bubbles that can be > >>> less than a > >>> > micron across. Ours is a regular (not superresolution) system > >>> - either > >>> > widefield fluorescence, or confocal fluorescence, or > >>> (hopefully, in the > >>> > near future) a Fluoview 3000 with deconvolution. Has anyone > >>> proved that > >>> > accurate measurements on such a small scale are possible? > >>> What > >> would > >>> it > >>> > take? Some kind of deconvolution - 2D? 3D? Does anyone know > >>> what others > >>> > have done in this direction? Thank you > >>> > > >>> > Mike Model > >>> > > >>> > >>> > >>> > |
«
Return to Confocal Microscopy List
|
1 view|%1 views
| Free forum by Nabble | Edit this page |