A rapid detection of early-stage cervical cancer using nucleus and cytoplasm-based machine learning-driven refractive...

Cervical cancer is one of the most common cancers in women around the globe and death from such cancer is largely due to late or insufficient diagnosis. Traditional forms of screening like Pap smears and HPV testing, though common with extensive use, have limitations in sensitivity, false negative and require repeated clinical follow-ups. To overcome these diagnostic gaps, we have presented a supportive layout of a Bisected Circular-Resonator and Axial Line Refractive Index Biosensor (BCALRIB) with a machine learning-driven approach that can help to detect cervical cancerous cells in the early stage. The recommended layout of geometry has emerged as especially valuable because cancer-related cells present greater nuclei and changed nuclear-to-cytoplasmic ratios, which produce distinctive optical signals. Additionally, an acceptable tolerance analysis has been evaluated to check and enhance the device's robustness against deviations in biological components. The optimum impressive sensitivity values of 1000.00 nm/RIU, 933.33 nm/RIU, and impressive detection limit values of 0.008289 RIU, 0.006935 RIU have been obtained for cervical cancerous nucleus (CCNu), and cervical cancerous cytoplasm (CCCy), respectively, with reference to healthy nucleus (HeNu), and healthy cytoplasm (HeCy). The impressive value of 175.81 quality factor has been obtained. The impressive machine learning parameter, resulting in a 0.996351 R-squared score with 6.057260 × 10^-05 of mean square error score. The current biosensors with SPR provide rapid, specific, and real-time detection of cervical cancer biomarkers by utilizing the variation of refractive indices on the sensor interface.