The use of light patterns to improve optical resolution was first proposed in 1963 by Lukosz and Marchand. The practical implementation of this theory was pioneered by Tony Wilson in the 1990’s and further developed by John Sedat, Mats Gustaffson and Reiner Heintzmann at the turn of the millennium. For more information about the history of the technique see here: http://www.microscopyu.com/references/superresolution/sim.html
In SIM a patterned grid is placed in the light path and the sample is imaged using different phases of light and grid positions. The striped pattern is carefully devised so that when light waves interact with the sample they produce interference patterns termed Moiré fringes. The captured interference images contain high frequency spatial information about the specimens.
By correlating the Moire Patterns generated when varying the rotation and phase of the grid with the known resolution limit of the system (termed optical transfer function) it is possible to mathematically restore the sub-resolution spatial information about the sample. The image on the left shows the raw data image set acquired using three different rotations and five different phases on the ESRIC N-SIM system. The right hand side image is a SIM reconstructed dataset. SIM can be carried out in 2D and 3D and is an excellent methodology for capturing the whole of a cell or organelle volume using multiple fluorophores.
An ideal SIM image is a sample is perfectly flat, has excellent contrast between the grid patterns, termed modulation contrast and no background noise. However in the real world of biology this rarely happens. So to optimise SIM image processing the SIM processing algorithms use Weiner filters to minimise noise in the output image and other filters to compensate for less than ideal contrast in the pattern modulation.
Sample preparation for SIM is in some respects as straightforwards as it is for confocal. It is important to use high quality precision thickness coverslips and to have optimal immunofluorescent staining using bright dyes, e.g. AlexaFluors and curing mounting medium. For antibodies or samples which are inherently more ‘dirty’ e.g. none specific antibody staining or autofluorescence, the background noise generated can interfere with the SIM reconstruction algorithms which will reduce the resolution. The improvement in resolution for SIM in 1.7 fold better than Abbe limited methods e.g. confocal. The ESRIC system has 4 laser lines and the best resolution expected is shown in the table below.
|N-SIM Laser (nm)||Typical dye used||Resolution Limit (nm)|