issues with non-blinking weak spots

The very bright, round foci clearly do not correspond to objects in the conventional image and are probably all artificats from non-switching foci.

These dots start off at moderate single molecule brightness (can’t eliminate them just using higher threshold cutoff without loosing lots of ligit switching events), slowly bleach and flicker above and below the threshold for detection, but never really disappear.

Not easily rejected by link filter due to threshold (could drop threshold but maybe that creates other problems, certainly slows the fitting down a lot just to fit some dots you plan to throw out.  Maybe screen the 256×256 image pixel by pixel for bright non-changing pixels and just toss those in all frames.

Potential solution might be to create a threshold mask with the conventional image.  But in denser images we’ll grab more false dots.  Want something like the intensity of the conventional to be related to the intensity / number of localizations.

Filtering by tossing pixels with low standard deviation (first 5000) frames, provides a somewhat tighter filter.  Still requires loading the original dax file, which is not incredibly fast nor currently accessible option from STORMrender…

 High laser power avoids these issues by bleaching all the weak nonswitching dots faster… STORM4 unique problem at present.  Maybe 642 with better dichroic match will work better.
This entry was posted in General STORM and tagged , . Bookmark the permalink.