Spherical aberration cannot be simultaneously corrected across all wavelengths and lowers image SNR. (A–D) γ-Enhanced PSF xz views of the same multicolor fluorescent microsphere (TetraSpeck; Thermo Fisher Scientific) imaged at different emission wavelengths (528/48 nm and 683/40 nm) using different refractive index immersion oils. (A and D) Optimized PSFs are symmetric in z, with confined central maxima. (B and C) Spherical aberration results in axial asymmetry and broadening of the PSF. xy images (insets in B and D) show reduced signal intensity collected at the focal plane resulting from spherical aberration (linear intensity pseudocoloring heat map displayed to the right).
The effect of large probe size and low labeling density. (A) The relatively large size of antibodies can both cause steric hindrances that reduce labeling density and increase the distance between the fluorophore and the structure of interest. (B) Smaller probes may allow for higher labeling density and bring fluorophores closer to the structure of interest. (C) Inadequate labeling density can result in an ambiguous or misleading speckled appearance for structures that are actually continuous (D).
Common problems in SRM images. (A and B) SIM reconstructions of a standard slide (Argolight SIM) that contain continuous line structures. (A) A SIM image reconstructed from high SNR raw images closely matches the expected structures. (B) A SIM image reconstructed from low SNR raw images containing artifacts, including the appearance of structures in the background and curved lines and discontinuities, within expected structures (arrow), which were not present in the sample. For display, negative intensity values were set to 0, and images were autoscaled. (C and D) SMLM images of Alexa Fluor 647–labeled microtubules created from datasets with lower (C) or higher (D) numbers of diffraction-limited spots that contain multiple simultaneously emitting fluorophores. The high-density dataset used to create the image in D was simulated by summing the first and second halves of the dataset in C to double the effective number of emitters in each raw image while keeping the total number of emission events constant. Molecules were localized using ThunderSTORM (single-emitter algorithm; Ovesný et al., 2014) and visualized as normalized Gaussians, with 10-nm pixel size. The presence of multiple emitters within a diffraction-limited area degrades resolution in areas of the specimen that contain higher fluorophore density, such as areas where structures overlap (circled areas). Notably, this effect is less pronounced in sparse curvilinear objects (arrows); estimating resolution using a line scan across this microtubule would misrepresent the resolution achieved elsewhere in the image.