PhD Candidate, CREOL


Non-imaging characterization by statistical analysis of scattered light under random illumination

Computational optical imaging and sensing are rather new multidisciplinary fields in which one try to take advantage of the available computation power, both in hardware and software, in conjunction with optical instruments to enhance certain attributes of the optical characterization processes. Following this philosophy, we proposed a computational method for characterising objects’ length scales. For instance, when we talk about a cell culture, length scales define averaged size of the alive cells. These length scales and their dynamic over the time encode specific biological processes. However, since these activities are slow and happen statistically, to follow them, one has to investigate them over an extended time that results in overall high level of light exposure energy. Unfortunately, the high exposed energy is shown to have collateral damages, e.g. changing the natural activity of the cell during the measurement that can result in the erroneous measurement. For example, it is shown that the visible light illumination to embryos in a specific level, even for less than 30 minutes, can stop their development. Here we are after addressing this issue through developing a sensing method in which one can measure the length scale (spatial information) over the time (dynamic information) without imaging the target. Our method is based on illuminating the target with randomly structured light, with parametric statistical properties, and then measuring the scattered light power only using a single pixel power meter. To avoid the requirement of the high illumination level, we don’t use either indigenous or exogenous tags. Having scattered light power for different illumination statistical parameters, we show that one can infer spatial properties through statistical analysis of the measured power. The time resolution of the method is 3 orders of magnitude faster than usual biological activities time scales (hundreds of milliseconds). Moreover, since we don’t use any tag and we rely on only measured power, the whole process can be done in an extremely low light condition, i.e. 3 orders of magnitude less than usual white light microscope illumination level. These properties, in conjunction with simplicity and versatility of the method, makes it an excellent candidate for the time-lapse biological study.