What question did this study set out to answer?

To develop a biosensor for early detection of cervical cancer by utilizing optical properties and neural network classification.

February 28, 2026Open Access

Enhanced design of a photonic crystal biosensor for early cervical cancer detection using refractive index variation and neural network classification

Key Points

To develop a biosensor for early detection of cervical cancer by utilizing optical properties and neural network classification.
Designed an optical sensor using a grid of silicon rods that guides light.
Tested sensor performance across four different temperatures: 10 °C, 25 °C, 45 °C, and 60 °C.
Utilized a neural network to classify tissue types based on optical features identified by the sensor.
Sensor demonstrated over 91% accuracy in classifying healthy, HPV-infected, and cancerous tissues.
Performed reliably across different temperatures, with optimal performance at 25 °C.
Showed high sensitivity to detect small variations in tissue properties.

Abstract

This study introduces a new optical sensor designed to help detect cervical cancer early, using light and special materials called photonic crystals. The sensor has been built using a simple grid of tiny silicon rods that guide light through a small pathway. When this light passes through different types of cervical tissue (healthy, HPV-infected, or cancerous) it changes slightly, de-pending on the tissue’s properties. These changes help to identify the type of tissue without needing any dyes or labels. A key innovation of the proposed work is considering the impact of temperature on sensor performance, while previous studies have ignored temperature effects. Since temperature can affect how tissue interacts with light, we tested the sensor at four different temperatures (10 °C, 25 °C, 45 °C, and 60 °C) to make sure it works well in different environments. The sensor works best at 25 °C, but it still performs reliably at other temperatures. The results show that this sensor is highly accurate, sensitive, and can detect even tiny differences in tissue. To enhance the classification of tissue types, an artificial neural network (ANN) was trained using key optical features from the sensor, achieving over 91% accuracy. It also has a simple design that makes it easier to build and use in real-world medical settings. Overall, this sensor has strong potential for non-invasive, fast, and reliable cervical cancer screening.

Mark Helpful

Bookmark

Relay

View Full Paper