Abstract
In Coulter counters, cells counting and volumetry is achieved by monitoring their electrical print when they flow through a polarized micro-orifice.However, the volume measurement may be impaired when the trajectory of the cell is in the vicinity of the aperture edges due to complex dynamics and deformations of the cell.In this thesis, numerical simulations of the dynamics and electrical signature of red blood cells (RBCs) in a Coulter counter are presented, accounting for the deformability of the cells.In particular, a specific numerical pipeline is developed to overcome the challenge of the multi-scale nature of the problem.It consists in segmenting the whole computation of the cell dynamics and electrical response in a series of dedicated computations, with a saving of one order of magnitude in computational time.This numerical pipeline is used with rigid spheres and deformable red blood cells in an industrial Coulter counter geometry and compared with experimental measurements.The simulations not only reproduce electrical signatures typical of those measured experimentally, but also provide an understanding of the key mechanisms at play in the complex signatures induced by RBCs following a near-wall trajectory.Based on this new understanding provided by numerical simulations, a filtering strategy is introduced, which allows the filtering of pulses induced by near-wall paths which are irrelevant for the cells sizing.The method is shown to retrieve the expected symmetrical distribution of RBCs and provides results comparable to hydrodynamical focusing, a more intricate implementation of the Coulter principle.Such a result paves the way for a robust assessment of haematological parameters with a cheaper and simpler implementation, compared to hydrofocused devices.The impact of the cell morphology and rheology on the electrical print is evidenced for near-wall trajectories.Indeed, by altering the cell deformability and sphericity, the electrical pulses are proven to differ from predefined normality of measurements.Furthermore, neural network modellings are performed in the aims of assessing such RBC properties.Among the proposed processing, classification of normal, stiffened and spherical RBCs is provided.Finally, the inverse problem of numerical simulations is achieved, thus allowing the evaluation of the mechanical parameters of RBCs.