3D Tracking Algorithms For Digital Holographic Microscopy And Its Applications In Monitoring Bacterial Behaviors

Swimming bacteria have complicated interactions with their environment and among themselves.The typical interactions include hydrodynamic interactions between cell-surface or cell-cell,steric repulsion between cells,and quorum sensing effects based on biological adaptation etc.Real-time tracking of the 3D dynamic behaviors of individual bacteria provides us a direct way for understanding these interactions,which is of great significance for understanding the mechanisms of disease transmission,plant diseases,and Marine fouling.Digital holographic microscope（DHM）has the advantages of large depth of field,high resolution,and label-free.It is very suitable for real-time,in-situ 3D observation of bacteria.The key to applying this technique to bacterial interaction studies lies in the 3D tracking algorithm.Although the tracking of swimming bacteria has been realized to analyze bacteria-surface interactions,it is still challenging to monitor two bacteria in close proximity as a bacteria pair.The image overlap of neighboring bacteria prevents the continuous tracking of an approaching bacterial pairs.The goal of this dissertation is to develop a 3D tracking algorithm to arm DHM for observing the 3D encounter process of a swimming bacteria pair,so that quantitative analysis of the bacteria-bacteria interplay can be realized.This dissertation introduced 3D tracking algorithm based on iterative search and another algorithm based on image recognition,respectively.The performances and limitations of the two methods were discussed.Using these methods,we investigated the encounter of E.coli pairs at different concentrations.The results revealed the individual heterogeneities leads to significant changes of kinetic energy during encounter,and the hydrodynamic interaction dominant the encounter dynamics thus improves the efficiency in exploring the surrounding environments.The following are the main contents of this dissertation:（1）A 3D tracking algorithm based on iterative search was developed for 3D simultaneous tracking of encountering E.coli pairs.The impact of bacteria encounter to the3D tracking of DHM was firstly assessed.3D trajectories were frequently interrupted,so that the observation of the interplay between cells became difficult.To address this problem,we iteratively reduce the searching radius and intensity threshold,allowing us to track the bacteria approaching in 3D.This algorithm could identify and recover the gap between 3D trajectory segments raising by the interruption from other bacteria through lateral image recognition and axial localization.At bacterial densities lower than 10⁸ m L^-1,the correct rate of the interrupted trajectories（rate of correctly linked trajectories）is close to 100%.The limitations of this algorithm are discussed at the end of this section,namely the long-time cost and the inability to distinguish bacteria in extremely close proximity（<1.5μm）.（2）To overcome the limitations of the iterative search algorithms,a 3D tracking algorithm based on image recognition was developed.First,the corresponding trajectory segments of pairwise bacteria were firstly identified from all the segments due to temporal and spatial correspondence.The focused image was reconstructed in the coordinates-missing area.A 2D intensity threshold algorithm was applied to the reconstructed image to yield the2D position of the object bacteria cell.Finally,the z coordinates are acquired from the Gaussian fitting of the 3D light field.The correct rate of this method（chance of stably tracking a bacterium to move forward one frame）is better than 95%when the bacterial density is less than 10⁸ m L^-1.Compared to the iterative search algorithm,it is much faster and can distinguish bacteria in extreme close proximity（<1.5μm）.We discussed the application scope of the two methods and find that they could complement well with each other.（3）Attributed to the bacterial growth and the surface accumulation effect,swimming bacteria like E.coli frequently approach with each other,especially when they swim near a surface.With the help of DHM and the above tracking algorithms,we recovered 3D pairwise trajectories of E.coli at 5×10⁶,10⁷ and 3×10⁸ m L^-1 which facilitate quantitative analysis of bacterial interaction dynamics.We found that even at a comparable low density(5×10⁶ m L^-1),face-to-face approach and collision induces transient significant variation of kinetic energy and suppressed tumbling.The kinetic energy increased with the rise of the approaching angle of the two bacteria,and decreased with their closest distance.Slight variations in the kinetic energy changes were found in among E.coli pairs at the three concentrations.Single-cell simulation and tracking of mutants lack adaptive responses both suggest the dominant role of physical hydrodynamic coupling between two coordinated cells in a range of 10 body length.In summary,we developed two algorithms for DHM to track approaching bacterial pairs,so that we can study the pairwise bacterial interactions at various densities.Our findings reveal that hydrodynamic interactions play vital role in this process and show impacts to their adaptation to surrounding environments.Our results revealed the importance of interactions between bacteria in dilute suspensions,therefore shed lights on bacterial physiology and have wide implications on active particle-driven microdevices.