A novel backlight fiber optical probe and image algorithms for real time size-shape analysis during crystallization

Process analytical technology requires not only process suitable sensors, but also novel data processing approaches in order to make real time analysis feasible. In this paper, a novel in-line probe was designed, fabricated and tested for crystallization monitoring. The design benefits from state of the art optics and camera, fiber backlight illumination, and an optimized depth of focus and field of view. Image analysis steps to study both crystal size and shape are presented. These image analysis algorithms do not require manual-thresholding of individual images or time zero image subtraction, due to the use of a 'rolling ball' self-adapting background correction step. The approach is tolerant to blank images, noise, blurriness, out of focus objects, and common spatial or intensity variations. The method developed should help in the identification of changes in size and shape in crystal populations. Examples are presented for glass sphere standards, the crystallization of d-mannitol and l-glutamic acid, as well as an engineered needle-sphere mixture. A data binning strategy useful for future studies is also reported. The ultimate goal is control of crystallization under the Process Analytical Technology framework.