Integration of computational optics and machine learning for optimized SPR-based carcinoembryonic antigen detection

Carcinoembryonic antigen (CEA) is an effective biomarker for diagnosing and tracking cases of liver cancer, breast cancer, and colorectal cancer. In this study, a hybrid SPR biosensor combined with black phosphorus and MgO/Cu/MgO multi-layer is designed and developed to fully utilize the capabilities of black phosphorus in confining electric fields and increasing sensor sensitivity. The brute force algorithm is utilized to optimize sensor parameters. The accuracy of artificial neural network and adaptive neuro-fuzzy inference system models in simulating sensor responses is thoroughly validated. The designed biosensor has a sensitivity of 409.02 deg/RIU, a figure of merit of 132.25, and a quality factor of 146.65 RIU -1 at a wavelength of 633 nm, which has immense application opportunities for precise measurement of CEA.