Analysis of dynamic biological systems imagery

Abstract

Biological systems pose considerable challenges when attempting to isolate experimental variables of interest and obtain viable data. Developments in image analysis algorithms and techniques allow for further mathematical interpretation, model integration, and even model optimization ('training'). We formulate two distinct methods for obtaining robust quantitative data from time-series imagery of two biological systems: Paenibacillus dendritiformis bacterial colonies, and human gastric organoids. Boundary parameterizations of P. dendritiformis are extracted from timelapse image sequences displaying colony repulsion, and are subsequently used to 'train' a previously developed nonlocal PDE model through the means of error minimization between observation and simulation. Particle tracking is conducted for small colloidal beads embedded within human gastric organoids, and then used to perform particle tracking analysis. This information is analyzed to quantify the local complex viscoelastic properties of organoids' interior mucosal environment.