deletions | additions       diff --git a/MFA.tex b/MFA.tex  index 844005b..67c5204 100644  --- a/MFA.tex  +++ b/MFA.tex 
...
\subsection{Overlapping events}  In traditional photon counting imagingand centroiding with simple one-iteration algorithms,  it is usually ensured that the frames do not contain images of photons that have any overlap; overlap. However, in biological imaging with super-resolution microscopy the image acquisition speed is a critical parameter which can be shortened by imaging as many molecules as possible in each frame. While the separation of overlapping events is not possible with  simple one-iteration center-of-mass centre-of-mass algorithms,  algorithmsare not capable of guessing which proportion of the detected intensity in a pixel  that contains fit a PSF to the event can potentially resolve  overlapping intensity from more than one photon belongs events by fitting multiple PSFs  to which photon event. a region containing overlapping events.  However, in biological imaging with super-resolution microscopy the image acquisition speed is a critical parameter which can be shortened by imaging as many molecules as possible in each frame. The separation of overlapping events is possible with algorithms that fit several point-spread functions to an area containing overlapping events. ThunderSTORM's option for Multi-emitter Fitting Analysis (MFA) produces excellent results with recognising and separating overlapping EBCCD photon events, as shown in Fig~\ref{fig_mfa}.\textit{Use the cell data here?}  Cell data: around 460 event/frame, USAF 150 events/frame...     From RSI paper:\\  In our setup with a 100$\times$ objective and no extra magnification in the detection path, each 24$\times$24 μm pixel in the CCD corresponds to 160~nm in the sample plane. This satisfies the  Nyquist sampling limit for 320~nm resolution, a bit above the theoretical resolution limit of $\sim$220~nm for this wavelength and objective. The centroided image shows resolution improvement compared to the sum and brightest pixel images. However, considering the thermal drift associated with the long acquisition time due to the low EBCCD frame rate, the biased distribution of the centroid positions in the horizontal direction, and the many photons required for the centroided image, centroiding with sub-pixel accuracy is of limited benefit in this case. However, for EB-sensors with faster frame rates, centroiding could be used to improve the resolution of the final image.