Automatic multicell identification using a compact lensless single and double random phase encoding system.

We investigate the use of compact, lensless, single random phase encoding (SRPE) and double random phase encoding (DRPE) systems for automatic cell identification when multiple cells, either of the same or mixed classes, are in the field of view. A microscope glass slide containing the sample is inputted into the single or double random phase encoding system, which is then illuminated by a coherent or partially coherent light source generating a unique opto-biological signature (OBS) that is captured by an image sensor. Statistical features such as mean, standard deviation, skewness, kurtosis, entropy, and Pearson's correlation coefficient are extracted from the OBSs and used for cell identification with the random forest classifier. With the exception of the correlation coefficient, all features were extracted in both the spatial and frequency domains. Experiments are performed with single random phase encoding and double random phase encoding, and system analysis is presented to show the robustness and classification accuracy of the random phase encoding cell identification systems. The proposed systems are compact, as they are lensless and do not have spatial frequency bandwidth limitations due to the numerical aperture of a microscope objective lens. We demonstrate that cell identification is possible using both the SRPE and DRPE systems. While DRPE systems have been extensively used for image encryption, to the best of our knowledge, this is the first report on using DRPE for automated cell identification.