AI-Enhanced Photonic Biosensor Could Help Detect Glioblastoma Tissue

Source: AZoSensors, by Muhammad Osama (September 29, 2026). AI-generated summary by biochip.com, published October 1, 2026. Not independently reviewed.

A proposed photonic biosensor paired with an artificial neural network could distinguish modeled glioblastoma tissue from nearby peritumoral brain tissue, researchers reported in Scientific Reports. The ultra-compact device uses light to detect tiny differences in refractive index, a measure of how a material bends light, rather than requiring stains or fluorescent labels. Numerical simulations produced an optical sensitivity of about 933 nm/RIU and classified eight simulated tissue profiles with 100% accuracy under group-aware cross-validation. The findings are preliminary, but they point toward a small optical component that could eventually help surgeons identify brain tumor margins.

A Light-Based Tissue Sensor

Glioblastoma is an aggressive brain tumor that can infiltrate surrounding tissue, making its edges difficult to identify during surgery. Imaging methods such as magnetic resonance imaging and computed tomography can help guide care, but they may not provide rapid, real-time information at the surgical margin.

The proposed sensor uses a two-dimensional photonic crystal, a carefully patterned material that guides and controls light much as a row of barriers can steer water waves. It consists of a hexagonal array of silicon rods suspended in air, with 17 × 16 rods, a 600 nm lattice constant, and standard rod radii of 120 nm.

How the Device Detects Tissue

The researchers introduced intentional gaps, called defects, into the photonic-crystal lattice to form input and output waveguides and a central ring resonator. That ring, containing 11 dielectric rods with an optimized radius of 144 nm, acts as the sensing region.

A single-mode laser sends light at a wavelength of 1.55 μm through an optical fiber and into the device. When tissue sits over the sensing area, its refractive index shifts the wavelength at which the ring resonates. In practical terms, the sensor looks for a subtle change in light behavior that reflects a difference in the tissue it contacts.

Modeled Performance and AI Classification

Using finite-difference time-domain simulations in RSoft software, the team modeled light interactions with tissue at 25 °C. The system differentiated a modeled refractive index of 1.341 for pre-tumoral tissue from 1.344 for tumor tissue.

The simulations yielded a maximum quality factor of 4,777, a figure of merit of 2,333 RIU-1, and a detection limit of 4.2 × 10-5. The proposed sensor occupies just 76 μm2, a footprint small enough to be considered for integration into a lab-on-a-chip system or compact diagnostic probe.

The team also trained a feedforward artificial neural network, a type of software that learns patterns from examples, using ambient temperature, resonance wavelength, and peak transmission intensity. On eight baseline simulated profiles made with synthetic training data, the model achieved 100% classification accuracy under group-aware cross-validation.

Temperature Is a Practical Challenge

Temperature can affect refractive index and therefore move the sensor's resonance peaks. Across simulated temperatures from 25 °C to 40 °C, the peaks shifted, but the modeled responses for peritumoral and cancerous tissue remained distinguishable.

That result matters because a surgical sensing tool would need to function under changing conditions rather than only in a tightly controlled laboratory setting. Still, thermal stability will require further testing once a physical device is built.

Why This Matters

A label-free optical probe could potentially give surgeons faster information about whether tissue near a tumor margin resembles cancerous or surrounding tissue. The device's small size and reliance on refractive-index sensing could make it compatible with miniaturized diagnostic platforms.

For now, this is a computational design, not a validated surgical instrument. The next steps are fabrication and tests with real tissue samples in ex vivo and clinical settings, where researchers can determine whether the simulated performance holds up in the far more complex conditions of brain tumor surgery.

Topics: Biosensors · Brain & neurological health · Cancer

Start here: Biosensors explained