High-performance one-dimensional defective photonic crystal biosensor featuring semiconductor defect layer and machine learning classification for early cancer diagnosis

This study presents a highly sensitive one-dimensional defective photonic crystal (PC) biosensor designed for early cancer cell detection by monitoring shifts in resonant modes caused by refractive index differences between healthy and malignant cells within bio-cavity layers. The sensor design features alternating layers of SrTiO3 and bio-cavities, with a central defect and is evaluated using the transfer matrix method (TMM) to fine-tune parameters such as the thickness of the defect layer, the angle of incidence and the number of periods. Substituting the SrTiO3 defect layer with SiO2 improves the interaction between light and matter, boosting sensitivity from 650 nm/RIU to 700 nm/RIU. Further refinement results in a maximum sensitivity of 930 nm/RIU at an incidence angle of 60°, along with a quality factor of 5.03 × 105 using 20 layer pairs. The sensor achieves a low detection limit of 4.132 × 10−7 RIU, allowing for the detection of minor refractive index variations linked to different cancerous cell types. Moreover, machine learning models, such as Support Vector Machine and Random Forest, were developed to classify cells based on spectral wavelength and refractive index data, with Random Forest achieving a perfect accuracy of 100%. This proposed biosensor provides a cost-effective, accurate and swift method for biomedical cancer detection, showing great promise for early diagnosis and related bio-sensing applications.