Low-Cost Continuous-Wave Diffusive Microtomography with Fiber-Scanned White-Light Illumination
Alexander Ingold · 2026
Read Low-Cost Continuous-Wave Diffusive Microtomography with Fiber-Scanned White-Light Illumination by Alexander Ingold (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-25183.txt.
Introduction
Three dimensional imaging inside specimens usually needs costly kit: pulsed lasers, fast detectors, or X ray sources. Alexander Ingold presents an ultra low cost alternative, a continuous wave diffusive tomography system built from a smartphone microscope, a white LED fed into an optical fibre, and 3D printed micropositioners.
The fibre is scanned to vary the illumination, and a physics based forward model, optimised with machine learning, turns the measurements into full colour volumetric reconstructions. Demonstrations cover a scattering phantom, a tartrazine cleared poplar section, and fungal mycelium growing near an Arabidopsis root. A thick poplar branch is imaged with a side emitting fibre inserted into it. Ingold is candid that the findings are early and qualitative, but they show scanned fibre illumination on cheap hardware can yield useful three dimensional outputs.
Catalog record
| CREATOR | Alexander Ingold |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.25183
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Low-Cost Continuous-Wave Diffusive Microtomography with Fiber-Scanned White-Light Illumination Alexander Ingold Abstract Tomographic microscopy enables three-dimensional internal imaging but often requires expensive optical or X-ray instrumentation. Here we present an ultra-low-cost continuous-wave diffusive tomography (CWDT) system for biological samples. The system uses a smartphone microscope, a white LED coupled into an optical fiber, 3D-printed micropositioners, and a physics-based forward model optimized with machine learning. We demonstrate full-color volumetric reconstructions from a tartrazine-cleared poplar section, a scattering phantom, fungal mycelium near an Arabidopsis root, and thick poplar branch imaging with an inserted side-emitting fiber. The current results are qualitative and exploratory, but they show that scanned fiber illumination and inexpensive hardware can produce useful three-dimensional reconstruction outputs for low-cost microscopy experiments. 1 Introduction Tomographic microscopy, which enables 3D internal imaging of specimens, is a powerful but often prohibitively expensive capability. Key modalities such as optical coherence tomography and two-photon microscopy rely on costly components including ultrafast pulsed lasers and high-speed detectors. Similarly, micro X-ray computed tomography requires specialized X-ray sources and sensitive detector arrays. This capital investment limits use cases such as long-term field deployment, point-of-care medical diagnostics in low-resource settings, and hobbyist microscopy. Our goal is to create an ultra-low-cost micro-tomography system for imaging 3D volumes in biological samples. We utilize a novel machine-learning-based variation of continuous wave diffusive tomography (CWDT) [ 9 ] . Our primary innovations are the use of a physics-based forward model optimized with machine learning, combined with white-light illumination and a scattering-reduction dye to enable high-resolution, full-color volumetric imaging with inexpensive hardware. CWDT involves positioning a continuous wave, fiber-guided, LED light source in an array of positions behind the sample and imaging at each position. CWDT has been explored for breast cancer diagnosis [ 3 ] and functional neuro-imaging [ 8 ] . While many CWDT projects use near-infrared illumination to avoid the absorbance of hemoglobin in animal tissues, we chose white-light illumination because it aligns with the low-cost goal of the project and enables imaging and reconstruction of three-color volumes. To improve image quality, the sample is soaked in a strongly absorbing tartrazine dye, which can decrease light scattering within tissue [ 19 ] . This non-toxic, food-safe dye shifts the real and complex refractive-index components. The real part helps match tissue interfaces to reduce scattering, while the imaginary part increases absorption to suppress multiple scattering events, improving the signal-to-noise ratio and reducing the ill-posedness of the inverse problem. The resulting stack of illuminations is processed into a 3D physics and ray-tracing-based forward model of the tissue. Light is simulated to radiate from a single illumination point and travel through approximately 16 absorbance and scattering filters. We imaged a 940 \mu m x 940 \mu m x 250 \mu m volume of a tissue section with a mean squared error of 0.002. We modeled 20 depth slices for a z-axis sampling interval of 10 \mu m, an x/y-axis resolution of 1.5 \mu m on the surface, and approximately 50–100 \mu m resolution at the deepest depth. To demonstrate the versatility of this technique beyond prepared tissue sections, we applied it to the challenging task of in-vivo imaging of thick, living plant matter. For thick poplar branches, the illumination source is inserted into the sample rather than positioned beneath a thin section. The optical fiber is ground at a 45-degree angle, and the end is treated with reflective paint or metal sputter coating. The mirrored fiber directs the light out of the end of the fiber at a 90-degree angle [ 4 ] . This protocol enables imaging of living poplar branches still attached to the tree. Imaging thick samples while maintaining high resolution using optical fiber insertion could also be relevant to tomographic imaging of precancerous skin lesions. The overall CWDT image-formation model is summarized in Figure 1 . Figure 1: A simulated physics-based tissue model with learned absorbance parameters and a simulated light source approximates the sample images with multiple incident light source locations. 2 Methods 2.1 Microscope Diffusive tomography requires a movable transmission light source and a camera imaging the surface of the sample. The light source consisted of a white LED flashlight (BLITZU Gator 32) and an optical fiber (Thorlabs FG050LGA, 50 \mu m core, 125 \mu m cladding). One end of the optical fiber was positioned in direct contact with the LED. The other end of the optical fiber was positioned underneath thin sections or inserted into thick samples. Both ends of the fiber were positioned with TriAxis 3D-printed micropositioners [ 14 , 13 , 12 ] . The microscope was a low-cost 120x smartphone-based microscope. A 3D-printed frame supported the microscope, smartphone (Motorola ThinkPhone 50 MP camera), microscope slide, and two 3D-printed micropositioners for the light source in inserted or transmission positions. The light source was moved in 100 \mu m increments between image captures. Camera settings were adjusted for each sample, but focus, ISO, exposure time, and white balance were held constant across each image stack. 2.2 Optical Fiber Preparation For transmission imaging, the optical fiber was cut with scissors. For thick-sample imaging, the end of the optical fiber was modified so that light exits perpendicular to the fiber axis. The fiber was ground at a 45-degree angle with lapping sheets (Thorlabs LF5P Silicon Carbide 5 \mu m Grit). For rapid grinding, a lapping-sheet disk was mounted on a rotary grinder (Dremel 1.2A), and the fiber was held against the lapping sheet with a 3D-printed micropositioner for 10 minutes. Lapping can also be performed by hand. After lapping and inspection for defects with the miniature microscope, the end of the fiber was dipped in mirror-finish paint (Rust-Oleum Mirror Effect Spray Paint). This paint consists of reflective aluminum microparticles and a clear binder. Three to five dip coats were applied for complete coverage. 2.3 Poplar Sample Preparation Branch cuttings from one-year-old Populus nigra x deltoides trees were potted in Sunshine Mix #4 (Sun Gro Horticulture) and maintained in a greenhouse for two years under a 14 h light / 10 h dark photoperiod. Imaging samples were prepared from branches at varying developmental stages, specifically between the 5th and 10th internodes. A narrow strip of bark ( \sim 3 cm in length, \sim 200 \mu m in depth) was carefully removed using a miniature wood plane (Veritas Miniature Bench Plane, Lee Valley Tools) to expose the underlying xylem. For thick-sample imaging, the branch was clipped and used directly. For thin-section imaging, a second pass with the sample plane removed a \sim 200 \mu m thick tissue section. A 0.38 molar tartrazine solution (FD&C Yellow 5, sciencekitstore.com) was prepared by dissolving 10 grams of tartrazine in 50 ml of distilled water. Stirring and heating to 50 degrees Celsius were required to complete the dissolution. For tartrazine staining, samples were soaked overnight for 18 hours, patted dry, and mounted on a slide. Branch samples were fixed to the slide with quick-cure two-part epoxy, and the incision was covered with a thin layer of glycerol to preserve tissue clarity and prevent air infiltration. Thin sections were mounted on the slide with a coverslip and glycerol. 2.4 Phantom Preparation A phantom was prepared for testing the resolution of suspended particles at various depths and sizes.