We demonstrate the training of a generative adversarial network (GAN) for the prediction of optical property maps (scattering and absorption) using spatial frequency domain imaging (SFDI) image data sets that are generated synthetically with a free open-source 3D modelling and rendering software, Blender. The flexibility of Blender is exploited to simulate 5 models with real-life relevance to clinical SFDI of diseased tissue: flat samples containing a single material, flat samples containing 2 materials, flat samples containing 3 materials, flat samples with spheroidal tumours and cylindrical samples with spheroidal tumours. The last case is particularly relevant as it represents wide-field imaging inside a tubular organ e.g. the gastro-intestinal tract. In all 5 scenarios we show the GAN provides an accurate reconstruction of the optical properties from single SFDI images with a mean normalised error ranging from 1.0-1.2% for absorption and 1.1%-1.2% for scattering, resulting in visually improved contrast for tumour spheroid structures. This compares favourably with the ∼10% absorption error and ∼10% scattering error achieved using GANs on experimental SFDI data. Next, we perform a bi-directional cross-validation of our synthetically-trained GAN, retrained with 90% synthetic and 10% experimental data to encourage domain transfer, with a GAN trained fully on experimental data and observe visually accurate results with an error of 6.3%-10.3% for absorption and 6.6%-11.9% for scattering. Our synthetically trained GAN is therefore highly relevant to real experimental samples but provides the significant added benefits of large training datasets, perfect ground-truths and the ability to test realistic imaging geometries, e.g. inside cylinders, for which no conventional single-shot demodulation algorithms exist. In the future, we expect that the application of techniques such as domain adaptation or training on hybrid real-synthetic datasets will create a powerful tool for fast, accurate production of optical property maps for real clinical imaging systems.
SignificanceThere is a need for a cost-effective, quantitative imaging tool that can be deployed endoscopically to better detect early stage gastrointestinal cancers. Spatial frequency domain imaging (SFDI) is a low-cost imaging technique that produces near-real time, quantitative maps of absorption and reduced scattering coefficients, but most implementations are bulky and suitable only for use outside the body.AimWe aim to develop an ultra-miniature SFDI system comprising an optical fiber array (diameter 0.125 mm) and a micro camera (1×1 mm package) to displace conventionally bulky components, in particular, the projector.ApproachFirst, we fabricated a prototype with an outer diameter of 3 mm, although the individual component dimensions could permit future packaging to a 90% are feasible for detecting simulated squamous cell carcinoma.ConclusionsThis device shows promise as a cost-effective, quantitative imaging tool to detect variations in optical absorption and scattering as indicators of cancer.
Spatial frequency domain imaging (SFDI) is a low-cost imaging technique that maps absorption and reduced scattering coefficients, offering improved contrast for important tissue structures such as tumours. Practical SFDI systems must cope with various imaging geometries including imaging planar samples ex vivo, imaging inside tubular lumen in vivo e.g. for endoscopy, and measuring tumours or polyps of varying morphology. There is a need for a design and simulation tool to accelerate design of new SFDI systems and simulate realistic performance under these scenarios. We present such a system implemented using open-source 3D design and ray-tracing software Blender that simulates media with realistic absorption and scattering in a wide range of geometries. By using Blender’s Cycles ray-tracing engine, our system simulates effects such as varying lighting, refractive index changes, non-normal incidence, specular reflections and shadows, enabling realistic evaluation of new designs. We first demonstrate quantitative agreement between Monte-Carlo simulated absorption and reduced scattering coefficients with those simulated from our Blender system, achieving 16% discrepancy in absorption coefficient and 18% in reduced scattering coefficient. However, we then show that using an empirically derived look-up table the errors reduce to 1% and 0.7% respectively. Next, we simulate SFDI mapping of absorption, scattering and shape for simulated tumour spheroids, demonstrating enhanced contrast. Finally we demonstrate SFDI mapping inside a tubular lumen, which highlighted a important design insight: custom look-up tables must be generated for different longitudinal sections of the lumen. With this approach we achieved 2% absorption error and 2% scattering error. We anticipate our simulation system will aid in the design of novel SFDI systems for key biomedical applications.
This paper presents the development and validation of a cost-effective 3D-printed conductive phantom for EEG sensing system validation that achieves 85% cost reduction (£48.10 vs. £300–£500) and 48-hour fabrication time while providing consistent electrical properties suitable for standardized electrode testing. The phantom was fabricated using conductive PLA filament in a two-component design with a conductive upper section and a non-conductive base for structural support. Comprehensive validation employed three complementary approaches: DC resistance measurements (821–1502 ), complex impedance spectroscopy at 100 Hz across anatomical regions (3.01–6.4 k with capacitive behavior), and 8-channel EEG system testing (5–11 k impedance range). The electrical characterization revealed spatial heterogeneity and consistent electrical properties suitable for comparative electrode evaluation and EEG sensing system validation applications. To establish context, we analyzed six existing phantom technologies including commercial injection-molded phantoms, saline solutions, hydrogels, silicone models, textile-based alternatives, and multi-material implementations. This analysis identifies critical accessibility barriers in current technologies, particularly cost constraints (£5000–20,000 tooling) and extended production timelines that limit widespread adoption. The validated 3D-printed phantom addresses these limitations while providing appropriate electrical properties for standardized EEG electrode testing. The demonstrated compatibility with clinical EEG acquisition systems establishes the phantom’s suitability for electrode performance evaluation and multi-channel system validation as a standardized testing platform, ultimately contributing to democratized access to EEG sensing system validation capabilities for broader research communities.
Objectives Upper GI endoscopies are aerosol generating procedures (AGPs), increasing risk of spreading airborne pathogens. We aim to quantify mitigation of airborne particles via improved ventilation, specifically laminar flow theatres and portable HEPA filters, during and after upper GI endoscopies. Methods This observational study included patients undergoing routine oral gastroscopy in a standard endoscopy room with 15-17 air changes per hour, a standard endoscopy room with portable HEPA filtration unit, and a laminar flow theatre with 300 air changes per hour. A particle counter (diameter range 0.3μm-25μm) took measurements 10cm from the mouth. Three analyses were performed: whole procedure particle counts, event-based counts and air clearance estimation using post-procedure counts. Results Compared to a standard endoscopy room, for whole procedures we observe a 28.5x reduction in particle counts in laminar flow (p5μm for oral extubation (12.2x,p<0.01), reduction in particles <5μm for coughing/gagging (6.9x, p<0.05) and reduction for all sizes in anaesthetic throat spray (8.4x,p<0.01) but no significant effect of HEPA filtration. However, we find that in the fallow period between procedures HEPA filtration reduces particle clearance times by 40%. Conclusions Laminar flow theatres are highly effective at dispersing aerosols immediately after production and should be considered for high-risk cases where patients are actively infectious or supply of PPE is limited. Portable HEPA filers can safely reduce fallow time between procedures by 40%.
We demonstrate training of a Generative Adversarial Network (GAN) for prediction of optical property maps (scattering and absorption) using spatial frequency domain imaging (SFDI) image data sets generated synthetically with free open-source 3D modelling and rendering software, Blender. The flexibility of Blender is exploited to simulate 3 models with real-life relevance to clinical SFDI of diseased tissue: flat samples, flat samples with spheroidal tumours and cylindrical samples with spheroidal tumours representing imaging inside a tubular organ e.g. the gastro-intestinal tract. In all 3 scenarios we show the GAN provides accurate reconstruction of optical properties from single SFDI images with mean normalised error ranging from 1-1.2% for absorption and 0.7-1.2% for scattering, resulting in visually improved contrast for tumour spheroid structures. This compares favourably with 25% absorption error and 10% scattering error achieved using GANs on experimental SFDI data. However, some of this improvement is due to lower noise and availability of perfect ground truths so we therefore cross-validate our synthetically-trained GAN with a GAN trained on experimental data and observe visually accurate results with error of <40% for absorption and <25% for scattering, due largely to the presence of spatial frequency mismatch artefacts. Our synthetically trained GAN is therefore highly relevant to real experimental samples, but provides significant added benefits of large training datasets, perfect ground-truths and the ability to test realistic imaging geometries, e.g. inside cylinders, for which no conventional single-shot demodulation algorithms exist. In future we expect that application of techniques such as domain adaptation or training on hybrid real-synthetic datasets will create a powerful tool for fast, accurate production of optical property maps from real clinical imaging systems.