Axial Localization Of Digital Holographic Microscopy And Its Applications

Digital holographic microscopy（DHM）is a novel optical technique capable for real-time3D observation of multiple micron-sized objects with irregular shapes or textures.It is label-free,wide in depth of field and high-throughput,thus facilitates in-depth understanding of 3D dynamics of colloids,microorganisms and cells.However,the localization accuracy and requirements,especially in the axial direction differ a lot for colloids and microrganisms due to their distinct 3D motion features.For in-line DHM,numeric reconstruction of 3D optical fields for the holograms captured by DHM is generally needed to locate all the objects in 3D appear in the field of view.The observation depth（the axial range）is usually tens to hundredsμms with an axial localization accuracy of around 1μm or above.Therefore,to perform accurate and rapid axial tracking of colloidal particles,microorganisms and cells down to single-cell level,axial localization methods,i.e.,Gaussian fitting algorithm based on numerical reconstruction and localization algorithms based on image processing of the holograms（light intensity and interference ring size）were proposed.The main factors affecting the accuracy of these algorithms were also discussed.Based on these methods,we further applied DHM to locate E.coli in 3D and monitor the 3D migration of human spermatozoa.These works shed lights on accurate and robust 3D localization and tracking for DHM and expand its application.The main contents of this dissertation are as follows.（1）The numerical reconstruction based Gaussian axial fitting algorithm was proposed and successfully applied in locating nanoparticles and bacteria.The main factors affecting its axial localization accuracy were also discussed.Firstly,holograms and background images of polystyrene microspheres（PLPs）and E.coli were recorded at various defocus distances（d）.Numerical reconstructions of the object field were then performed by a convolution method.Subsequently,3D locations of particles（x,y,z）were determined by searching the local intensity maxima.z coordinates were then calibrated and the root mean square deviations between the calibrated z and d were calculated,i.e.the localization accuracy.Localization accuracy of the algorithm at various incident light intensity,illumination uniformity and objective magnification before and after introducing the Gaussian fitting refinement algorithm were assessed.The results indicate incident light intensity,illumination uniformity and objective magnification are the main factors affecting localization accuracy,while the Gaussian fitting refinement algorithm effectively improves the axial localization accuracy in all these recording conditions.As a result,the axial localization accuracy is refined down to58 nm for 0.2μm PLPs and 318 nm for E.coli,respectively.（2）Aiming to save computing time and improve the accuracy for observing near-surface dynamics,axial localization algorithm based on the intensity distribution of holograms was introduced and the main factors affecting its accuracy were discussed.Firstly,lateral coordinates（x,y）of the particles were determined by a centroid algorithm for holograms of PLPs.Intensity profiles were plotted against d and fit by a sinusoidal function to determine the working curve.Axial locations of a particle were determined from the working curve.Localization accuracy at various incident intensities and wavelengths were assessed.It is revealed that the axial localization accuracy is improved with higher incident intensity and shorter wavelength.Our results show as high as 4 nm axial localization accuracy over 230 nm working range can be successfully obtained.（3）For observing systems containing objects with different shapes,i.e.,bacteria cells from different species,a localization algorithm based on the diameter of the interference ring in the hologram was proposed.Holograms of PLPs and E.coli were recorded at various defocus distances（d）and normalized by background images.The diameter of the first order dark ring（w₀）was selected for axial localization.It was extracted from the intensity profiles at where the first order deviation of the intensity reach 0.The working curve of w₀to d was then determined and applied for determining z.Localization accuracy was then determined at various recording conditions.For E.coli,their orientation should be determined before the localization was performed.Recording conditions and sample textures were shown to be slightly affect on localization accuracy which is better than 590 nm over tens ofμms.（4）Finally,3D dynamics and migration of human spermatozoa near surfaces with various topographies were explored utilizing DHM.Surfaces with similar chemistry but tuned topographies were prepared by spin coating a single-layer of silica particles to the coverslide.Hologram sequences for human spermatozoa swim near the surfaces were recorded.By numerical reconstruction and 3D localization,their 3D trajectories were acquired.Motility parameters,such as 3D velocity,swimming orientation and motion patterns,and density profiles of human spermatozoa were then assessed.Results show microscale topographic surfaces modulate 3D dynamics of human spermatozoa through hydrodynamic interactions.Particularly,human spermatozoa resist this modulation by their tail beating at the topographic surface where the diameter of the coated silica particles is equal to the sperm tail width.