Polarized Light and Optical Angular Momentum for Biomedical Diagnostics 2022

Sensitive vectorial imaging, such as through polarization microscopes, requires both high spatial resolution and high-fidelty polarimetry. We present new methods to enhance such approaches. We show new polarimetry that maps all analyser states into a single vectorially structured light field, permitting single-shot vectorial state analysis through physical model based inference. These methods avoid complex processing steps and error sources. We further introduce a new adaptive optics technology that can correct both phase and polarisation aberrations within optical systems. We validate improvements in both vector field state and the focal quality, through correction for commonplace sources of vectorial aberration.

Interaction of light with extended biological tissues involves continuous changes in its coherence and polarization states. Therefore, the classic Stokes-Mueller calculus based on the local (single-point) light transformation on the order of intensity (not field) cannot be used for complete and unique reconstruction of the optical properties of random, birefrigent bio-tissues. We suggest a generalization of the Stokes-Mueller calculus to two-point field correlations in which both the Stokes vector and the Mueller operator remain real-valued. We show in particular that the classic Poincare construction enjoyed by the single-point polarization state can be extended to two points to characterize, in addition to the polarization state, its degree of electromagnetic coherence and its degree of cross-polarization. The two-point Stokes-Mueller calculus can be used in a number of applications, e.g. spatial light modulator analysis but is envisioned to be of particular interest in biophotonics.

Imaging Mueller polarimetry (IMP) was used in reflection geometry and large field of view configuration for the in-plane visualization of brain fiber tracts by exploring the anisotropy of the refractive index of healthy brain white matter. Our initial studies demonstrated that IMP successfully detects in-plane orientation of fiber tracts on a flat surface of the excised brain specimens. This work, performed ex-vivo on complete fresh calf brains proves the potential of IMP as a technique suitable to detect both presence and orientation of brain fiber tracts in the adverse conditions of complex surface topography and presence of blood.

Alignment of tissue fibers, such as collagen, can be disrupted by some cancers. Mueller matrix polarimetry (MMP) is often correlated to this tissue orientation, as seen in cases such as cervical imaging. However, its bulk assessment of a sample’s polarimetric properties prevents it from distinguishing specific structures at different depths, as there would be in any non-uniform or multilayered sample. Spatial Frequency Domain Imaging (SFDI) is a well-documented technique that can be used to manipulate depth of penetration of an investigating light source through use of different sinusoidal frequency patterns. We have combined the two imaging modalities to investigate and differentiate between anisotropic samples with varied depth-dependent structure and demonstrate its use with various tissue phantoms. We also discuss some issues that arise with the combined methodology.

Polarimetry has been used to characterize collagen fiber alignment in biological tissues, providing insights into structure-function relationships during unique loading conditions or in disease. These imaging modalities leverage the birefringence of collagen to describe fiber strength and orientation. Structural anisotropy is thought to be the primary driver of this optical phenomenon; however, the degree to which other extracellular matrix (ECM) properties (i.e., local density, crosslinking) confound polarimetry data has yet to be explored. In this study, the effects of individual ECM properties on polarized light propagation in transmission and diffuse reflectance modes was explored using a tunable collagen hydrogel system.

We propose a predictive method for biological tissue recognition based on the experimental Mueller matrix measurements of a collection of 157 ex-vivo chicken samples corresponding to four different tissue sets: bone, tendon, muscle and myotendinous junction. We highlight the potential of the predictive algorithm, built by means of the statistical analysis of different polarimetric observables that retrieve complete polarimetric information of samples (that is, dichroism, retardance and depolarization). Composed by four independent functions, the proposed predictive optical-based algorithm is capable to recognize in a single measurement, whether a random sample corresponds, or not, to one of the mentioned tissue groups.

Radiation concerns preclude the use of fluoroscopy guidance in lumbar puncture procedures performed on pregnant women or other at-risk groups. We have developed an alternative method for guidance based on a simple, unscanned polarization-sensitive optical coherence tomography needle tip probe. Using the porcine spine as a model, we show that the polarization signals returned by the probe allow each layer from the skin to the subarachnoid space to be uniquely identified in situ. Combining these signals with needle-tip tracking using Doppler methods provides real-time anatomical localization of the needle tip.

Reflectance imaging gives a convolved image of superficial and deeper tissue layers. On the other hand, polarization reflectance imaging allows separation of superficial tissue from the convolved tissue image. In this report, polarized light imaging is used to investigate hydration and desiccation on superficial layers of ex vivo porcine skin tissues. A polarization camera acquired co-polarized and cross-polarized reflectance images from porcine front and back dermal surfaces. Polarized scattering is sensitive to sub-micrometer changes in tissue structure, and therefore is useful in detecting collagen density changes in tissue. The reflectance images were acquired at five different wavelengths (405, 490, 590, 660, and 700 nm) for hydrated and desiccated tissues. The back surface (dermal surface) of the skin was affected by hydration or desiccation, while the front surface (epidermal surface) was not as affected due to the stratum corneum which resists desiccation.

We present a combination of Mueller matrix measurements (635 nm) of cancerous colon specimens and machine-learning approach. Physical realizability filtering and symmetric decomposition were used to extract polarimetric quantities, used as predictors in machine-learning algorithms. The results were visualized using various depolarization spaces. Principal component analysis was used to extract particular features from the model, logistic regression evaluated predictors with high likelihood for tumor detection, while random forest and support vector machines provided the best results for classification. Hence, polarimetry combined with machine-learning approach may increase the histopathology diagnostic accuracy.

This work describes a wide-field imaging Mueller polarimetric system optimized for operation in the NIR range. It makes use of two fast switching compensators with optimal values of retardance that are oriented to a discrete set of orientations. The acquisition of 16 raw intensity images is done in a few seconds, which allows the determination of the complete Mueller matrix of a sample in the backscattering configuration at a speed compatible with in-vivo applications. Mueller matrix NIR imaging of polarization properties unveils quantitative information from deeper parts of the tissue than in systems using visible radiation, making it an interesting and promising tool for non-invasive biomedical diagnostic.

Preterm birth (PTB), defined as any birth prior to 37 weeks of gestation, contributes to 35% of 3.1 million neonatal deaths annually. There is an absence of clinical tools available for accurate and early detection of preterm birth risks. We have developed a Portable PReterm IMaging System capable of 4x3 MMI which can be used at the point of care for monitoring of the cervix. The system was tested on a Gynecologic skills trainer as well as healthy volunteers showing image quality comparable to clinical colposcopes.

Thin sections of uterine cervices from pregnant mice at day 6 and 18 of 19-days gestation period at different spatial locations along the cervices were studied with imaging Mueller polarimetry (IMP) combined with statistical analysis and multi-curve fit. The results suggest using depolarization and linear retardance images for collagen scoring and identification of cervical collagen changes during pregnancy. One day before delivery the remodeling of extracellular matrix of cervical collagen was detectable at the external cervical os, thus, proving that IMP modality may serve for the preterm birth risk assessment associated with the accelerated remodeling of cervical collagen.

Mounting evidence supports the prognostic value contained within the tumour stromal architecture, more specifically the desmoplastic response (DR). However, despite its promise, DR suffers from inter-observer variability, making its clinical uptake difficult. In this study, we are applying our novel polarized light microscopy technique, combined with advanced texture analysis and machine learning algorithms, to offer a quantitative assessment of DR and to investigate the potential prognostic capabilities of our approach. The overarching goal of this study is to expand on our polarized light imaging + quantification methodology, as a step towards improved cancer stratification and personalized therapy management.

Polarized nonlinear imaging is an important tool to image molecular organization in tissues using intrinsic nonlinear or multiphoton contrasts. The dynamics of polarized nonlinear imaging is however today still limited to seconds time rates. We have developed an interferometric frequency shift-based method using Acousto-Optic Modulation, capable of reaching sub-second dynamics in polarized scanning imaging. The method is applied to real time structural imaging in cells and tissues.

Precancer-to-cancer progression in epithelial tissues is marked by cellular proliferation and nuclear pleomorphism. These transformations have been shown to influence tissue optical properties: (1) tissue turbidity (quantified by the scattering coefficient) and (2) average scatterer size. It is found that backscattered polarized light can assess these properties. We use linear and circular depolarization to generate a predictive response surface where turbid media samples cluster by scatterer size and scattering coefficient. The novel polarimetric ability to noninvasively and simultaneously assess scatterer size and scattering coefficient of tissue-like turbid media may improve upon currently invasive skin cancer tests.

A three-dimensional label-free multi-contrast imaging for ex vivo tissue investigation is presented. Computational refocusing is implemented in a Jones-matrix polarization-sensitive optical coherence tomography (PS-OCT) system to overcome the trade-off between imaging depth and lateral resolution. The application of multiple contrast imaging, including intensity, birefringence, and degree-of-polarization uniformity (DOPU), is demonstrated by phantom, porcine muscle, and zebrafish measurements. Extended imaging depth with enhanced lateral resolution over millimeter is achieved. In tissue imaging we find some altered birefringence and DOPU estimation, whose size and alteration are proportional to defocus amount. This biased estimation can be numerically mitigated after applying computational refocusing.

One million infants will die of complications from reduced gestation before the age of five. Monitoring elastin and collagen arrangement in the uterine cervix interests researchers studying preterm birth. We have utilized a Self-validating Mueller matrix micro-mesoscope (SAMMM) with convolutional neural networks and K-nearest neighbor for classification of the fibers in the mouse cervix. A polarized microscope can be used for fiber classification leveraging the developed classifier. The Mueller matrix and decomposition parameters from excised cervical tissues are fed to the previously developed classifier. This low cost tool, powered with machine learning, could improve access to collagen and elastin imaging.

The structural alterations in biological tissues influence the tissue optical properties such as scattering, birefringence and optical activity. We developed a Stokes vector polarimetric system to investigate the peculiarities of the phase change of circularly polarized light during its propagation through biotissue-mimicking phantoms and real biological tissues. With the help of tissue screening experiments, Monte-Carlo simulation, and the Poincaré sphere approach, we investigated the change of polarimetric properties of light backscattered from the specimens due to the variation in the measurement’s sampling volume, change in tissue scattering and the process of optical clearing.