Dahan Kim

Date of Award


Level of Access Assigned by Author

Campus-Only Thesis

Degree Name

Master of Science (MS)




Samuel Hess

Second Committee Member

R. Dean Astumian

Third Committee Member

Sharon Ashworth


Despite the resolution limit (~200nm) due to the diffraction of light, fluorescence microscopy has been a widely used method to delineate biological processes because of its ability to perform non-invasive imaging with molecular specificity in physiological conditions of living cells. The diffraction limit of far-field fluorescence microscopy has been overcome with the advent of a family of localization microscopy techniques including fluorescence photo-activation localization microscopy (FPALM), capable of laterally resolving objects within 20nm or smaller. In localization microscopy, simultaneous multi-color imaging of multiple fluorescent species has been affected by bleed-through because of the uncertainty in the probe classification. By assigning molecules of one species to the channel of another, bleed-through poses a significant challenge in obtaining an accurate super-resolution image by creating false colocalization, and in quantifying correlations with pair correlation or Pearson correlation coefficient analyses by artificially increasing correlations. We present a novel method of grid-based bleed-through correction which allows calculations of pair correlation and Pearson coefficients unaffected by these artifacts of bleed- through. By correcting for bleed-through on the basis of grids, this method can determine the correct number of each species within each grid element, even allowing grid-based rendering of super-resolution images corrected for the bleed-through. In addition, we introduce a phase- averaging method of Pearson coefficient calculations, which computes the correlation coefficient characteristic of the image itself, unaffected by the uncertainties associated with the relative position between the grid and the image. The results on the simulated data show our method of bleed-through correction successfully restores correct values of correlation under various rates of applied bleed-through. Furthermore, super-resolution images corrected for the bleed-through show an absence of false colocalization, which appears in images rendered without bleed-through correction. Using only bleed-through rates, this correction method can be applied in principle to multi-color imaging of any localization microscopy method that employs similar multi-species imaging techniques. Furthermore, apart from the bleed-through correction, the phase-averaged Pearson coefficient analysis can even be applied to co-distribution analysis in confocal microscopy. When applied to two-color imaging of influenza hemagglutinin and cytoskeletal p-actin, the method shows a positive correlation between the two species, indicating a confined movement of hemagglutinin on the cell membrane along the underlying cortical actin cytoskeleton. Corrections applied to two-color imaging of HA and cofilin reveal that there is an anti-correlation between the two, quantifications of which could not be obtained without using the bleed-through correction. Because of the desire for quantification of the interactions between biological molecules, and because of the sensitivity of these quantifications to bleed-through, this bleed- through correction method is expected to form an important resource for users of localization- based super-resolution microscopy.