Fig. 2: Comparative performance of MLAO methods in a 2-P microscope.
From: Universal adaptive optics for microscopy through embedded neural network control
a Residual aberration after one correction cycle for three methods. Points show the mean and the shaded area indicates the standard deviations (SDs) of aberration distributions. The images show an example field of view (FOV) when different amounts of a random aberration were introduced. b–f show the normalised intensity metric (yI) as a proxy for correction quality, against the number of images used for multiple iterations of correction when random aberrations were introduced. In (b), an ensemble of ten random aberrations were corrected, imaging over the same FOV. Error bars on the plot showed the SDs of the fluorescence intensity before and after correction. c–f show specific corrections imaging microtubules of BPAE cells, illustrating performance for different specimen structures and imaging conditions. The images were acquired before and after correction through the different methods (as marked on the metric plots). Insets on the images show residual wavefronts after correction for each image. The grayscale colorbars show phase in radians
