Artificial Intelligence in Digital Holographic Imaging

8No‐search Focus Prediction in DHM with Deep Learning

8.1 Introduction

DHM allows for nondestructive investigations of biological samples as well as marker‐free and time‐resolved studies of cell biological processes. More specifically, interpretation of quantitative phase signals with DHM gives access to quantitative measurements of both cellular morphology and sample content with only a single shot. It is a real‐time approach that can be used for time‐lapse studies of biological samples in the absence of a mechanical focus adjustment. The propagation distance must be determined to obtain a quantitative phase image for phase objects. The distance between the hologram plane (CCD plane) and the observation plane (image plane) is defined by the reconstruction distance d. In digital holographic reconstruction, an in‐focus image is reconstructed when the reconstruction distance is equal to the distance between the CCD plane and the image plane during the hologram recording (see Figure 8.1). An out‐of‐focus image appears if d is not precise (see Figure 8.2). Several automated approaches have been proposed to find the best focus plane in DHM [1–6]. Generally, multiple images at different focus planes are numerically reconstructed and a focus‐evaluation function determines whether the image is focused by assessing the sharpness of either the amplitude‐contrast or phase‐contrast image. For example, pure phase objects have minimum visibility in amplitude‐contrast images when in‐focus ...