Design of THz photonic crystal fiber based biosensor for detection of brain tissues and behavior characterization with Machine learning approach

In this research, we proposed a Terahertz (THz) refractive index-based Hollow-Core Photonic Crystal Fiber (HC-PCF) biosensor for examining various brain cancerous tissues. Six design variants with cladding segments ranging from 4 to 16 are analyzed using the finite element method (FEM). The biosensor demonstrates high sensitivity (94.9 to 97.46%), minimal Effective Mode Loss (EML) of 0.00246 cm −1 with an effective mode area of 2.84 × 10 −8 m 2 and a power core ranges from 93% to 95.9% for the 16-segment cladding. The second contribution involves applying machine learning (ML), utilizing Autoencoder Augmentation Network (AEAN) for data augmentation and Bayesian Ridge Regression Multioutput Regressor (BRRMOR) for rapid prediction of biosensing parameters. The effectiveness of the ML model is demonstrated with a high r 2 score of 0.992 for unknown HC-PCF structures, showcasing computational efficiency compared to Finite Element Method simulations.