ROI detection in SPR measurements and molecule binding parameter estimation

WCU Author/Contributor (non-WCU co-authors, if there are any, appear on document)
Le Chen (Creator)
Western Carolina University (WCU )
Web Site:
Yanjun Yan

Abstract: Since 1983 when Surface Plasmon Resonance (SPR) was first proposed, it has become a widelyusedmethodology for various biosensing applications. In a SPR biosensing system in arrayformat, locating the region of interest (ROI) and estimating themolecule binding parametersfrom the SPR measurements are of great importance. In this thesis, we addressed these twochallenges by detecting the ROIs automatically and estimating the parameters optimally throughthe minimization of the mean square error (MSE).We first pre-processed the SPR video frame images to help enhance the ROI detectionperformance, and then applied the randomized Hough transformto automatically detect theROIs. With hundreds or even thousands of ROIs on a single SPR video frame image, our procedureto automatically detect the ROIs greatly reduced the labor to assign the ROIs.We then extracted the image gray level intensity data fromthe ROIs as a function of time,which were used to estimate the molecule binding parameters, ka (the rate of association) andkd (the rate of dissociation). These binding parameters are vital in biosensing applications. Inthis research we use a Particle SwarmOptimization (PSO) algorithm to estimate the parametersand compared the performance to the commercially used Levenberg-Marquardt (LM) algorithm,a gradient based algorithm. Our PSO algorithm performed better than LM achieving a muchlower MSE for all the active ROIs.

Additional Information

Language: English
Date: 2015
circle detection, data fitting, image processing, particle swarm optimization, surface plasmon resonance
Surface plasmon resonance
Biosensors -- Technological innovations
Biosensors -- Mathematics

Email this document to