No abstract

No abstract

The imaging field of view (FOV) of lensless microscope is consistent with the size of image sensor in use, enabling the observation of sample areas larger than 20 mm2. Combined with high-performance and even super-resolution phase retrieval algorithms, micron and sub-micron resolution can be achieved, ultimately realizing wide-field and high-resolution imaging performance simultaneously. However, high-throughput lensless imaging poses significant challenges in terms of rapid data acquisition and large-scale phase retrieval. Additionally, when observing biological samples over a large FOV, the focus plane often exhibits inconsistency among different regions, necessitating further parameter calibration. In this study, we propose a fast acquisition and efficient reconstruction strategy for coherent lensless imaging based on a multi-height imaging model. Multiple measurements are manually modulated using an axial translation stage and continuously captured by an image sensor, facilitating rapid data acquisition within seconds and requiring no hardware synchronization. The efficiency and accuracy of phase retrieval are enhanced through precise parameter calibration algorithms, as well as techniques such as region-wise parallel computing and region-wise auto-focusing. Experimental results demonstrate 7.4*5.5 mm2 FOV and 1.55 um half-pitch resolution imaging of human skin and lung tumor sections with region-wise focusing, requiring only a 2-s acquisition time and approximate 44-s reconstruction time. Furthermore, by incorporating the pixel super-resolution principle, the 1.10 um half-pitch imaging resolution is demonstrated in full-FOV peripheral blood smears without additional data required, beneficial to the identification of hollow shape and segmentation of blood cells.