Bayesian Regularized Neural Network-Based SPR Biosensor for Early Cancer Detection

This paper investigates on biosensor utilizing surface plasmon resonance (SPR) mechanism, which provides improved sensitivity and capability for detecting cancer cells. The biosensor is composed of gold (Au) and molybdenum disulfide (MoS2) as the sensing layers. Finite Element Method (FEM) is utilized in numerical analysis along with wavelength-interrogation technique to detect changes in the refractive index (RI) of human cells both normal and cancerous. Using Bayesian Regularization Artificial Neural Network (BRANN) models, parameters like mean square error (MSE) and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> were evaluated to determine the efficiency of the biosensor. The best MSE and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values were obtained for the number of neurons (N) 45 i.e., 13.727482 and 0.999830, respectively. It can be used on more complicated structures to observe parameter correlation, cover a wide range of parameters, and achieve more accurate findings, especially with a high number of inputs and less time and computing effort. This biosensor exhibited a peak wavelength sensitivity (WS), quality factor (QF) and resolution (R) of 31428.57 nm/RIU, 106373.63 nm/RIU and 3.182×10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-6</sup> RIU for MCF-7 cell. Additionally, figure of merit (FOM) and detection accuracy (DA) of 271.43 RIU<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> for MDAMB-231 cell and 13.33 μm-1 for HeLa and Jurkat cells. The suggested technique has a lot of promise to enhance sensing capacity.