Quantitative Zernike Phase-Contrast Microscopy with an Untrained Neural Network

• DOI: 10.1364/jsapo.2024.16p_a37_3
• Publication Date: 2024-12-24T17:12:49Z
• Authors: Zinan Zhou; Keiichiro Toda; Rikimaru Kurata; Kohki Horie; Ryoichi Horisaki; Takuro Ideguchi

In optical microscopy, Zernike phase-contrast microscopy (PCM) is a technique that transforms phase shifts in a sample to contrast in intensity by interference. Despite its wide usage in many biological and clinical applications, it fails to provide quantitative information about the specimen. One prior collaborative work [1] from our group managed to add quantitativeness to PCM by a phase retrieval algorithm based on compressive propagation. However, this algorithm relies heavily on regularization and non-trivial optimization tricks, severely limiting its generalizability and usage in practical situations.