Circularly polarized light scattering imaging of a cancerous layer creeping under a healthy layer for the diagnosis of early-stage cervical cancer

Significance: Cervical cancer progresses through cervical intraepithelial neoplasia (CIN), which are precursor lesions of cervical cancer. In low-grade CIN, atypical cells generate inside the squamous epithelium, which causes the accuracy of cytodiagnosis for cervical cancer not to be very high. The grade of CIN can be estimated by the depth of atypical cell infiltration from the basal layer to the surface, rather than the abnormality of cells. Therefore, a non-invasive method is required to evaluate the depths of abnormal cells hidden at depths. Aim: Cancerous tissues beneath healthy tissues were experimentally identified by using circularly polarized light scattering (CiPLS). This method enabled the changes in the size of the cell nuclei within the penetration depth in tissue to be investigated. Approach: Artificial unexposed cancerous tissues were prepared that consisted of healthy/cancerous/healthy layers with various thicknesses of the topmost healthy layer and the cancerous layer. A polarization imaging camera with a quarter-wave plate was used to create distribution images of the circular polarization of the scattered light. Results: CiPLS images indicated that the thickness variation of the top healthy layer (the depth of the cancerous layer) caused significant changes in the degree of circular polarization. Conclusions: The depth of unexposed cancer lying within the optical penetration depth can be evaluated using a circular polarization imaging system based on the CiPLS method. These findings will lead to the development of a non-invasive optical diagnostic method for early-stage cervical cancer, potentially improving early detection and treatment outcomes.

💡 Research Summary

Cervical cancer remains a leading cause of mortality among women worldwide, with most cases arising from persistent human papillomavirus (HPV) infection that progresses through cervical intra‑epithelial neoplasia (CIN). The grade of CIN is defined by the proportion of atypical cells occupying the squamous epithelium, which in turn reflects how deep these abnormal cells have infiltrated from the basal layer toward the surface. Low‑grade CIN (CIN 1–2) often evades conventional cytology because the atypical cells are not exposed on the surface, leading to sub‑optimal diagnostic accuracy. Consequently, a non‑invasive method capable of probing several hundred micrometers beneath the epithelial surface is highly desirable.

The present study investigates circularly polarized light scattering (CiPLS) as a means to detect cancerous tissue hidden beneath a healthy epithelial layer. In the Mie‑scattering regime, where scatterers (cell nuclei) are larger than the illumination wavelength, the degree of depolarization of polarized light depends strongly on the size ratio between wavelength and particle diameter. Circularly polarized light (CPL) retains its handedness far better than linearly polarized light after multiple scattering events, because forward scattering dominates and the rotation of the polarization plane is minimal. Therefore, CPL that penetrates into the tissue, interacts with nuclei, and returns to the surface carries information about the size distribution of the scatterers within a few millimeters of depth.

To test this principle, the authors fabricated artificial multilayer samples consisting of a top healthy layer, a buried cancerous layer, and a bottom healthy layer. The cancerous layer (1 mm thick) was made from human pancreatic cancer xenograft tissue, while the healthy layers were derived from mouse hind‑limb muscle. The top healthy layer thickness was varied (0, 0.5, 1.0, 1.5 mm) to simulate different depths of a hidden lesion. The total sample thickness exceeded 3 mm, ensuring that the near‑infrared illumination (617 nm and 850 nm) would experience significant multiple scattering but still retain measurable polarization.

The optical setup employed two high‑power LEDs (617 nm, 1 W; 850 nm, 1.6 W) whose output was converted to right‑handed CPL using a linear polarizer, a half‑wave plate, and a wavelength‑specific quarter‑wave plate (QWP). The CPL beam illuminated the sample at ±30° incidence, producing an elliptical spot on the surface (major axis ≈ 29 mm, minor axis ≈ 25 mm). A polarization‑sensitive camera (Toshiba Polarsens, 8 MP) equipped with a 2 × 2 micro‑grid of wire‑grid polarizers (0°, 45°, 90°, 135°) captured four simultaneous intensity images. By processing these raw images, the Stokes parameters S₀–S₃ were reconstructed; the degree of circular polarization (DOCP) was defined as |S₃|/S₀. Because the camera does not provide a direct measurement of total intensity, the authors approximated S₀ with the square‑root of the sum of the four pixel intensities, a “pseudo‑intensity” that proved sufficient for relative DOCP comparisons.

Raw intensity images showed the biological tissue as bright regions surrounded by transparent agarose. DOCP maps revealed a clear dependence on the thickness of the overlying healthy layer. At 617 nm, increasing the top layer thickness caused a monotonic decrease in average DOCP, whereas at 850 nm the opposite trend (monotonic increase) was observed. This wavelength‑dependent behavior aligns with prior findings: shorter wavelengths are more strongly depolarized by smaller normal nuclei, while longer wavelengths are more sensitive to the larger nuclei characteristic of cancer cells. To suppress contributions from surface reflections, roughness, and other systematic artifacts, the authors computed ΔDOCP = DOCP₆₁₇ nm − DOCP₈₅₀ nm. ΔDOCP images displayed a clean contrast that correlated directly with the buried cancer layer depth, confirming that the differential circular polarization signal isolates the scattering contribution of the cancerous nuclei.

The experimental ΔDOCP trends matched Monte‑Carlo simulations previously reported by the same group, which predicted a detectable cancer depth of approximately 1.6 mm—well within the typical squamous epithelium thickness (~0.7 mm). Sensitivity analysis showed that even a 0.5 mm variation in the overlying healthy layer produced a statistically significant shift in ΔDOCP, indicating that the method can resolve sub‑millimeter depth changes.

While the results are promising, several limitations must be acknowledged. The healthy tissue used (mouse muscle) exhibits pronounced anisotropy due to its fibrous structure, potentially influencing the measured polarization beyond the intended Mie‑scattering effects. Human cervical epithelium is considerably thinner, more heterogeneous, and contains both columnar and squamous components, which may alter the optical path and scattering regime. Moreover, the current system relies on fixed incident angles and manual calibration of the QWP; real‑time clinical deployment would require robust alignment, automated intensity normalization, and possibly machine‑learning‑based depth extraction algorithms.

In conclusion, the study demonstrates that circularly polarized light scattering imaging can non‑invasively detect a cancerous layer concealed beneath a healthy tissue slab and quantify its depth within the optical penetration range of near‑infrared light. By exploiting the differential DOCP response at two wavelengths, the technique effectively isolates the scattering signature of enlarged cancer nuclei while suppressing surface artifacts. This proof‑of‑concept paves the way for developing a bedside, stain‑free optical diagnostic tool for early‑stage cervical cancer, potentially improving screening accuracy for low‑grade CIN lesions that are currently missed by conventional cytology. Future work should focus on translating the method to in‑vivo cervical imaging, optimizing system ergonomics, and validating diagnostic performance in clinical cohorts.