Fluorescence microscopy is characterized by low background noise, thus a fluorescent object appears as an area of high signal/noise. Thermal gradients may result in apparent motion of the object, leading to a blurred image. Here, we have developed an image processing methodology that may remove/reduce blur significantly for any type of microscopy. A total of ~100 images were acquired with a pixel size of 30 nm. The acquisition time for each image was approximately 1second. We can quantity the drift in X and Y using the sub pixel accuracy computed centroid location of an image object in each frame. We can measure drifts down to approximately 10 nm in size and a drift-compensated image can therefore be reconstructed on a grid of the same size using the “Shift and Add” approach leading to an image of identical size asthe individual image. We have also reconstructed the image using a 3 fold larger grid with a pixel size of 10 nm. The resulting images reveal details at the diffraction limit. In principle we can only compensate for inter-image drift – thus the drift that takes place during the acquisition time for the individual image is not corrected. We believe that our results are of general applicability in microscopy and other types of imaging. A prerequisite for our method is the presence of a trackable object in the image such as a cell nucleus.
Applications of Digital Image Processing, 2013
Main Research Area:
Progress in Biomedical Optics and Imaging
Imaging, Manipulation, and Analysis of Biomolecules, Cells, and Tissues, 2013
SPIE - International Society for Optical Engineering