Documentation of results table

[AnnWh]

In the documentation for ThunderStorm there doesn't appear to be specific documentation for what each column in the results table are. I have read several versions of the support documentation and can't find this information. Bearing in mind that several of our Super-resolution Microscopy facility users are Cell and Molecular Biologists by training with a very limited understanding of advanced Maths it would be helpful if this were more clear.

My reading is that there are the scalars for lateral positions (X and Y) following subpixel fitting. Sigma, which here is the accuracy of fitting (no description of how this is calculated in a easy to access place, presumably this does depend on how the subpixel fitting is done but this needs to be more clear).

Intensity the intensity of photoswitch in 'photons' (dependent on camera settings'). Offset, No idea what this actually is. If its the background reading it should be calculated per frame and not per localisation.

BGSTD. I would imagine this is Background Subtraction, however the acronym could mean anything.

Uncertainty. Is this a measure that combines the accuracy of fitting and also intensity. Or is it a derivative of sigma?

Thansk in advance for the calrification

[cleterrier]

Hi,

Sigma for a localization is the width if the actual spot in the image in nanometers. It is usually around 160-200 nm for STORM with Alexa Fluor 647.

BGSTD is the standard deviation of the local background in the localization fitting area (something like 5x5 pixels). It is used to calculate the uncertainty of the localization (error on X, Y and Z coordinates).

Uncertainty is the uncertainty of the localization. uncertainty_xy is for the lateral uncertainty, uncertainty_z for axial uncertainty. It is used when rendering the image with Gaussians at each localization: the width of the Gaussian corresponds to the uncertainty of the localization fitting.

Offset (I think) is the baseline level i.e. the constant background on top of which the localization signal is fitted. As it's a local value, there is a different background for each localization.

[AnnWh]

Hi Christophe

Awesome and thanks for putting this together for me. I've written a manual for my our core facility users and collaborators all of whom are Biologists and they're less familiar with some of the terminology used in the SMLM field so need a bit more granular explanation. Its very helpful to hear from the authors of the software exactly what is going on. I'd not realised the background was calculated from a 5x5 region I;d thought the average intensity of that particular frame was used - so thanks for explaining.

The explanation for Sigma is particularly useful as I can advise people to gate out anything above 200nm sigma if they have used AF647. This makes our end users lives a lot easier.

ThunderStorm is a fantastic tool and we all very much enjoy using it in ESRIC . We appreciate all of your work and your description here.

Ann

Dr Ann Wheeler

Head of IGMM Advanced Imaging Resource

IGMM, Edinburgh EH4 2XU, UK

---

From: Christophe Leterrier notifications@github.com Sent: 29 August 2016 14:06:54 To: zitmen/thunderstorm Cc: WHEELER Ann; Author Subject: Re: [zitmen/thunderstorm] Documentation of results table (#34)

Hi,

Sigma for a localization is the width if the actual spot in the image in nanometers. It is usually around 160-200 nm for STORM with Alexa Fluor 647.

BGSTD is the standard deviation of the local background in the localization fitting area (something like 5x5 pixels). It is used to calculate the uncertainty of the localization (error on X, Y and Z coordinates).

Uncertainty is the uncertainty of the localization. uncertainty_xy is for the lateral uncertainty, uncertainty_z for axial uncertainty. It is used when rendering the image with Gaussians at each localization: the width of the Gaussian corresponds to the uncertainty of the localization fitting.

Offset (I think) is the baseline level i.e. the constant background on top of which the localization signal is fitted. As it's a local value, there is a different background for each localization.

You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/zitmen/thunderstorm/issues/34#issuecomment-243118483, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ASfOyjdrZBqcr3y6BfIPbC5cUdAQ-r32ks5qktlugaJpZM4IfMJP.

The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

[cleterrier]

Hi Anne,

Happy to help! Actually I'm not an author of ThunderSTORM, more a regular (and happy) user. I made scripts to translate localizations file to and from ThunderSTORM format, so I had to dig into the meaning of the different columns. The 5x5 pixels is just an example (not sure if it is exactly 5x5, I think it depends on the settings) but it is definitely a very local window. About the Sigma, actually what you do when you restrict the size is make a virtual "optical sectioning" as you reject out-of-planes localizations that have a larger PSF (if you're not strictly in TIRF mode). You can play with the filtering on sigma and see how you vary the thickness of the images plane.

By the way I enjoyed your talk at SMLMS (I watched it on YouTube)!

Cheers,

Christophe

[kclwcic]

Hi Anne,

I just came across this searching for the same information. Would you be willing to share the manual you wrote - it would be very helpful to grasp the scope of such a document for guiding how I teach our users this software. Best Wishes, George Chennell