Exposure per frame

Submitted by s.wu on Fri, 2015-06-05 18:09

Forums

Unblur & Summovie

I'm a little confused by the parameter of "Exposure per frame". I initially thought that the exposure (e-/A^2) referred to the dose received by the specimen. But the method part of the elife paper (eLife 2015;10.7554/eLife.06980) says "The exposure per frame as reported by Digital Micrograph (Gatan, Inc.) was 0.769 e-/Å2 which corresponds to an exposure of 8 electrons/pixel/s on the camera." When I used 8e-/pix/s to calculate the total exposure, it gave me 100e-/A^2. This means the value of 100 corresponds to the dose on the camera. It's not clear if that's the dose received by the camera or 'counted' by the camera, which is lower because of the imperfect QE and coincidence loss. In the analysis of T20S proteasome, the exposure value appears to be the dose received by the camera, i.e. counts measured by the camera compensated by loss. According to the original NRAMM paper (eLife 2015;10.7554/eLife.06380): "at a dose rate of ∼ 9 counts/physicalpixel/s which corresponds to ∼12 electrons/physical pixel/s". The value of 12 gives the total exposure value of 53e-/A^2.

My questions are,
1. when I specify "Exposure per frame", shall I use the value readout by the camera, received by the camera, or received by the specimen?
2. the filter formula and constants are derived based on the beam intensity that gives 8 electrons/pixel/s on the camera. Do you think it is applicable to data collected using much lower/higher beam intensity?
3. Can we use the same exposure filter function for Falcon or DE data?

Alexis Fri, 2015-06-05 19:02

Hi,

Here would be my answers:

1. "Exposure" refers to the number of electrons incident on the specimen, per unit area
3. The filter formula concerns the damage of protein under electron exposure, it is independent of the detector.
2. I'm not aware of reports that describe protein damage due to electron exposure where the amount/rate of damage is a function of the exposure rate (number of electrons per unit time), so I would guess that the exposure filter applies regardless of beam
intensity. But others may have more knowledge of this matter than I do.

Hope this helps,
Alexis

timgrant Fri, 2015-06-05 21:24

Hi,

You raise a very interesting point here. We used a value of 100e-/A2 under the assumption it were the exposure that the specimen was exposed to. It is my understanding that the version of digital micrograph that we used corrected for the coincidence loss etc and thus the reported value is in (corrected) electrons and not detected counts (therefore we state 8 electrons per pixel in the paper - not counts). The critical exposure curves, and the dose filter curves are plotted under this assumption - i.e. that the viruses were exposed to 100e/A2 in total. We thus used a value of 53e-/A2 for the NRAMM proteosome data assuming this is the corrected value and thus an approximate estimate for the exposure of the specimen.

The intention is thus that you should be entering the true (or approximate) exposure that the specimen is exposed to, which would be camera independent.

I will double check this in case I am wrong - but I believe you should be using the exposure the specimen is exposed to, and the camera should not matter.

Thanks,

Tim

timgrant Tue, 2015-06-23 13:43

I have now double checked this with Chris Booth from Gatan - he confirmed that the adjustments are done internally by Digital Micrograph, thus the exposure values we quote in the paper should be approximately equal to the exposure the sample receives.