Unveiling Fourier Ptychography via Fisher Information

Abstract

In recent years, Fourier Ptychographic Microscopy (FPM) has been demonstrated for a wide variety of applications including wide field of view pathology and quantitative phase imaging. Briefly, this technique produces a high resolution reconstruction by computationally combining a set of diversely acquired captures of the same sample. However, the nonlinear aspects of the algorithm and the noise in a typical dataset affect the reconstruction quality. The ability to weight differentially each image within an entire FPM capture would allow to device an optimal design of an acquisition platform and to potentially avoid such reconstruction problems. Currently there is no method to calculate the information content of each capture and therefore to weight it accordingly. In this work, we propose such a method based on the concept of Fisher Information, not available until now for FPM acquisitions. This criterion depends on the selection of a sample model and the parameters to be measured in an FPM experiment. We show the application of our technique to two common problems: volume quantification using phase images and distance determination between two particles, both revealing interesting internal aspects of Ptychography. The approach used here could lead to an improvement of the speed in the acquisition and reconstruction process by making a more efficient selection of the illumination angles.