Vicena Logo Vicena
Optical simulation · Vicena Compute beta

Simulate light from source to sensor.

Explore tissue optics, photonics, spectral rendering, lenses, filters, cameras, and detector pipelines with open scientific engines—then inspect the fields, assumptions, and design tradeoffs.

Tissue light transport

Spectral optical systems

Sensor and camera pipelines

MCX / pmcx · executed result
850 nanometre light fluence simulated through layered tissue with GPU Monte Carlo

Example study

From invisible photons to a design you can inspect.

This real workflow starts with an everyday idea—a near-infrared wearable sensor—and progressively turns it into a bounded physical model, a GPU computation, visual evidence, and a defensible next experiment.

1,000,000

Photon histories

Bounded GPU Monte Carlo transport

850 nm

Near-infrared source

A common wearable-sensing wavelength

0.5 mm

Voxel resolution

60 × 60 × 40 tissue grid

RTX 4090

Managed GPU

MCX through pmcx 0.7.1

01

Design goal

Start with the sensing question

The study begins with an intuitive engineering question: how far does 850 nm light travel through layered tissue, and where could a wearable detector recover a useful signal?

02

Physical model

Turn anatomy into a model

Vicena converts layer thicknesses, absorption, scattering, anisotropy, refractive index, source position, and voxel size into an explicit MCX configuration.

03

GPU compute

Trace photons on the GPU

One million photon histories move through the voxel volume with Fresnel boundaries. The managed job keeps the engine version, seed, grid, and optical assumptions attached.

04

Visual evidence

Inspect where the light goes

The three-dimensional fluence field becomes centre-plane heatmaps, depth attenuation, target-depth summaries, and lateral-spread profiles instead of one opaque scalar.

05

Quality gate

Challenge the result

Penetration, absorbed fraction, boundary behavior, photon-count sensitivity, voxel sensitivity, and detector assumptions are separated from experimental validation.

06

Next design

Iterate the sensor geometry

Source–detector spacing, wavelength, tissue layers, optical properties, and photon budget can be changed while preserving the same reproducible workflow contract.

Real workflow artifacts

The simulation becomes a visual scientific record.

Every figure below comes from the executed MCX notebook. The three-dimensional result remains available behind the plots, so the agent can answer follow-up questions without rebuilding the science from a screenshot.

Traceable outputs
Log-scale 850 nanometre Monte Carlo light fluence through layered tissue

A million photon paths, made visible

The real MCX result shows the rapid change in fluence across a 30 × 30 × 20 mm tissue volume. The source, skin boundary, and depth markers remain connected to the executed configuration.

Linear and logarithmic depth-dependent light fluence curves through tissue

Depth attenuation

Linear and logarithmic views expose both the near-surface maximum and the orders-of-magnitude falloff that matters for detector placement.

Lateral light-fluence spread at several tissue depths

Lateral spread with depth

Profiles at 2, 5, 10, and 15 mm show how scattering broadens the illuminated region while the available fluence decreases.

Log-scale comparison of simulated fluence at selected tissue depths

Decision-ready depth comparison

The result can be reduced into an engineering comparison without losing the underlying three-dimensional array and notebook.

Scientific packages

The scientific packages behind light transport, photonics, and optical design.

  1. MCX / pmcx

    pmcx 0.7.1

    GPU photon transport

    Traces photon histories through voxelized media for tissue optics, fluence, absorption, and detector-placement studies.

  2. Meep

    1.34

    FDTD electromagnetics

    Solves time-domain Maxwell equations for photonic structures, transmission, resonances, fields, and device response.

  3. MPB

    Photonic band structures

    Computes electromagnetic modes and dispersion relations for periodic dielectric structures and photonic crystals.

  4. Mitsuba

    3.9

    Spectral rendering

    Models light transport through surfaces, materials, lenses, filters, cameras, and differentiable imaging systems.

  5. pymoo

    0.6.2

    Multi-objective optimization

    Explores optical design tradeoffs such as signal, absorption, geometry, robustness, and competing objectives.

  6. SALib

    1.5.2

    Sensitivity analysis

    Quantifies how uncertain optical properties and design variables influence simulated outputs and decisions.

Scientific capabilities

One optical workspace, several modeling layers.

Vicena routes the question to the appropriate open scientific engine and keeps model boundaries explicit—from scattering tissue to spectral surfaces and sensor pipelines.

Tissue light transport

Model absorption, scattering, anisotropy, refractive-index boundaries, sources, and voxelized media with GPU MCX.

Spectral optical systems

Use registered Mitsuba workflows for surface, lens, filter, spectral-rendering, and differentiable-imaging studies.

Sensor and camera pipelines

Connect simulated light to filters, quantum efficiency, image formation, calibration, noise, demosaicing, and downstream analysis.

Design exploration

Compare wavelengths, geometries, optical properties, source–detector spacing, robustness, and multi-objective tradeoffs.

Evidence discipline

A result you can question and extend.

The value is not a colorful heatmap by itself. It is the connection between the question, physical assumptions, managed execution, arrays, plots, provenance, and the next design decision.

The notebook records wavelength, grid, voxel size, optical coefficients, photon count, random seed, and boundary model.
Full fluence arrays remain available for new sections, regions of interest, and sensitivity analysis.
Plots are derived from the executed MCX output—not painted concept illustrations.
Engine and GPU provenance stay attached to the managed-compute result.

Scientific boundary

What this result does not prove.

! The optical coefficients are literature-reasonable assumptions, not measurements from a particular person or tissue phantom.
! This demonstration uses one fixed seed and one voxel size; consequential design needs repeatability and spatial-resolution studies.
! Fluence at a point is not a calibrated photodiode signal. Detector area, acceptance angle, spectral response, noise, and electronics must be modeled.
! MCX is a voxel-domain Monte Carlo engine. It does not replace full-wave electromagnetics or mesh-based MMC for every optical geometry.

Start with your design

What would you like to illuminate?

Start with a real optical design question. Vicena can select the workflow, prepare the model, run bounded compute, and return the visual evidence.