Associate Professor University of Arkansas, United States
Introduction:: The study of the tumor microenvironment and tumor angiogenesis is an emerging area of interest for research and clinical communities. Understanding tumor progression is essential to developing cancer treatments and therapies. Current methods for monitoring the tumor microenvironment are limited in their ability to access a wide range of anatomical locations, specifically the gastrointestinal tract, without causing non-physiologic tumor response. In this study, we aim to develop a small-diameter multi-projection endoscopic imaging platform capable of many different scientific and clinical applications, including assessing tumor structure and profusion in previously inaccessible anatomical locations. Due to the limited size available for an endoscopically compatible system, photon scattering follows a “transport regime,” (as opposed to the diffusion regime). This requires novel approaches to processing image data to resolve the fine vascular and structural details within tissue. The platform consists of two key components, one being a microendoscopic probe that utilizes radially spaced illumination source fibers to capture two-dimensional fluorescence images of simulated tissue. The other component, Monte-Carlo-based photon propagation simulations modeled after the colon epithelium, provides information in depth. The combination of these two components allows for the creation of a three-dimensional spatial reconstruction of a tumor embedded in a tissue of interest.
Materials and Methods:: An optical imaging probe (3.048mm diameter) containing eight radially spaced light projections (50mm ± 2% diameter) and a centralized imaging fiber (790 ± 50mm diameter) containing 30,000 ± 3,000 picture elements is used alongside a fiber-coupled LED with a wavelength of 530nm (Thorlabs Inc. M530F2) (Figure 1). Images are captured using an 8-bit CCD camera (FLIR Flea3 USB) with the FlyCapture2 platform. Purely scattering optical phantoms consisting of polystyrene microspheres (Polysciences Inc. 07310-15) were created to replicate the optical values of epithelial tissue in the colon. A reduced scattering coefficient of μs’ = 5cm-1 for ƛ = 532nm was selected. To simulate tumor microvasculature, a monofilament polypropylene suture (100μm diameter) (Redilene MP-3 6-0) was affixed in the phantoms and imaged at varying distances from the probe. The captured images are analyzed using MATLAB by sampling pixel values on various line profiles generated at arbitrary angles ranging from ±90° relative to the light source. By subtracting the resulting decay curves obtained from homogenous and nonhomogeneous phantom images, the presence and location of an embedded absorber can be determined (Figure 2). A characteristic Monte Carlo-based photon curve can then be selected for each generated decay curve. Photon curves are thresholded to select the regions of the curve with the highest photon transmission, creating a binary matrix (Figure 3). The matrix is then voxelized, generating a three-dimensional binary matrix (Figure 4). A three-dimensional reconstruction of the absorber can then be created using photon curves with varying degrees of rotation from the light source.
Results, Conclusions, and Discussions:: Homogenous and embedded absorber phantom images were captured with all eight illumination fibers at distances of 0μm, 50μm, and 101μm. Captured images were analyzed to identify distinct deviations in intensity values to determine where an absorber is located. There were quantifiable deviations from the homogenous case at corresponding distances where an absorber is present in the imaging field. A preliminary reconstruction of the absorber created from the combination of many Monte-Carlo photon simulations is shown in Figure 4. This study demonstrates a proof of concept for a minimally invasive method to quantify optical properties below the surface of epithelial tissue using a minimal diameter probe (< 1mm). The platform has shown to be a step in the right direction in the ability to identify the presence and location of an embedded absorber within a region of interest and then three-dimensionally reconstruct the identified absorber. By combining reconstruction data from all eight source fibers, a more accurate representation of the tumor, and its corresponding microvasculature, can be created. Further development of the platform’s imaging and reconstruction capability and accuracy is currently ongoing.
Acknowledgements (Optional): : This work was supported by the National Science Foundation (CBET 1751554), the National Institutes of Health, the Arkansas Integrative Metabolic Research Center (5P20GM139768-02), and the Arkansas Biosciences Institute.