A Fiber-Optic Surface Plasmon Resonance Sensor for Bio-Detection in Visible to Near-Infrared Images.

In this paper, we demonstrate a fiber-optic surface plasmon resonance (FO-SPR) biosensor based on image processing and back propagation (BP) neural network. The transmitted light of the FO-SPR sensor was captured by using visible (VIS) and near-infrared (NIR) CMOS sensors. The optical information related to the SPR effect was extracted from images based on grayscale conversion and an edge detection algorithm. To achieve accurate monitoring of refractive index (RI) changes, the grayscale means of the VIS and NIR images and the RGB summation of the edge-detected images were used as training and test inputs for the BP neural network. We verified the effectiveness and superiority of this sensing system by experiments on sodium chloride solution identification and protein binding detection. This work is promising for practical applications in standardized biochemical sensing.