AbstractsBiology & Animal Science

Biosensor arrays for molecular source identification in mass-transport systems

by Maryam Abolfath Beygi Dezfooli

Institution: University of British Columbia
Department: Electrical and Computer Engineering
Degree: PhD
Year: 2014
Record ID: 2043955
Full text PDF: http://hdl.handle.net/2429/45723


In this thesis, we investigate the use of biosensor arrays for detection of molecular species in dynamical fluid systems. First, we construct dynamical models for the transport of target molecules in a fluid over an array of surface-based biosensors. Using multiple time and length scales in the system dynamics, we develop a novel approximate model for the system. The derived model is an extension of the two-compartment model, which is used in the analysis of mass-transfer binding experiments. Then, we investigate how measurements of multiple biosensors can improve the estimation of target molecule concentration in a mass transfer system. The estimation problem is solved as a nonlinear least squares problem based on the derived model. The properties of the finite-sample estimator is investigated by deriving analytic expressions and Monte Carlo simulations. As an example, the results are illustrated for a protein-based biosensor. Second, we address the problem of detecting minute concentrations (nano to pico-molar) of target molecules with a single sensor. It is shown that substantial improvements in the response rate can be obtained by distributing the sensing surface area to form an array of spaced smaller sensors while the total sensing area remains fixed. The output signals of the individual sensors in the array are combined to form a single output signal for measurements. Formulas are derived for quantifying the improvement of the array response and for optimizing the size of the sensors. The analytical results are compared with experimental data for a protein-based biosensor and a surface plasmon resonance biosensor. Finally, we construct a mathematical proof for the accuracy of the two-compartment model, which is the basis of most of our analyses.