How far red light travels before tissue absorption and scattering steal its power
Photodynamic therapy (PDT) begins with a physical dosimetry decision: what wavelength, at what power density, delivered for how long. These three numbers — set entirely at the surface — determine everything about how deep the therapeutic effect can reach, because tissue optics are fixed by biology, not by the clinician.
Three device classes dominate clinical PDT delivery:
• Laser diode systems: narrow-linewidth, high-power, fiber-coupled — used for precise interstitial or endoscopic delivery through thin optical fibers • LED panels / arrays: broader-linewidth, lower cost, used for large-field superficial skin lesions (actinic keratosis, superficial BCC) • Cylindrical diffusing fibers: bare fibers with a scattering tip inserted directly into or adjacent to bulky tumors, radiating light radially outward — the main route to treating anything deeper than a few millimeters
Dosimetry separates two related quantities: fluence rate (instantaneous power density, mW/cm²) and fluence / dose (energy density integrated over treatment time, J/cm²). A field treated at 150 mW/cm² for 1000 seconds delivers 150 J/cm² — but the fluence rate, not the total dose, is what determines whether tissue at a given depth crosses the cytotoxic threshold at any moment.
Fluence rate and total dose are not interchangeable: doubling treatment time doubles the dose delivered at the surface, but it does nothing to extend the depth at which the instantaneous fluence rate exceeds the cytotoxic threshold — that depth is set purely by wavelength and starting power.
Surface illumination (a diode array or LED panel held above the skin, or a flat-cut fiber placed endoscopically against a mucosal lesion) is the default delivery geometry because it requires no invasive placement. Its simplicity comes at a direct optical cost: all photons must first cross the air-tissue interface and then travel straight into a highly scattering, moderately absorbing medium, losing fluence at every step.
Because the same wavelength/power inputs are used across all five stages of this model, the surface fluence rate and wavelength chosen here propagate through every downstream calculation — the scattering cloud, the absorption losses, the exponential decay curve, and ultimately the maximum treatable depth.
Once inside tissue, photons rarely travel in a straight line for more than a fraction of a millimeter. Elastic scattering off mitochondria, collagen fibrils, nuclei and membrane interfaces randomizes photon direction so efficiently that after a few scattering events, light transport is better described as diffusion than as a beam.
A photon entering tissue meets one of two fates at each interaction:
• Absorption: the photon's energy is captured by a chromophore (hemoglobin, melanin, water, or — usefully — the photosensitizer itself) and the photon ceases to exist as light. This is the event that ultimately drives the photodynamic reaction, but it also removes photons from the beam. • Scattering: the photon changes direction but keeps its energy. In soft tissue this is overwhelmingly elastic (Mie-regime) scattering off structures comparable to or larger than the wavelength — mitochondria (~1µm), collagen fibrils, nuclei, membrane interfaces.
In the 400–800nm range, scattering interactions outnumber absorption interactions by roughly 10–50 to 1 — tissue is optically "milky" rather than clear. Each individual scattering event is strongly forward-biased (anisotropy g ≈ 0.9, meaning photons prefer to keep moving roughly the same direction), but because thousands of such events accumulate over a few millimeters, net photon direction randomizes completely — the beam becomes a diffusing cloud.
The transport mean free path — the distance over which a photon's direction becomes effectively randomized — is given by ℓ* = 1/µs′, where µs′ = µs(1−g) is the reduced scattering coefficient. For typical soft tissue (µs′ ≈ 5–20 cm⁻¹), ℓ* works out to roughly 0.5–2 mm.
Beyond a few transport mean free paths from the source, the photon fluence distribution loses memory of the original beam direction and becomes accurately described by the diffusion approximation to the radiative transport equation — the same mathematical framework used for heat conduction. This is what allows the exponential (rather than purely geometric) decay law used in Stage 4, and it is why PDT dosimetry from a few millimeters down behaves like a diffuse glow rather than a beam, regardless of how tightly the source was originally focused.
Every wavelength competes against the same three dominant tissue chromophores — hemoglobin, melanin and water — but each absorbs light at wildly different rates depending on wavelength. The region where all three are simultaneously weak, roughly 600–1200nm, is the "therapeutic window" that makes deep-tissue PDT and photobiomodulation possible at all.
Three absorbers dominate visible-to-near-infrared tissue optics, each with a distinct spectral signature:
• Hemoglobin (oxy- and deoxy-): an intense Soret absorption band near 420nm makes blue light almost useless for anything but the most superficial treatment; smaller Q-bands persist at 540–580nm; absorption falls substantially by 600nm and keeps declining toward the red • Melanin: absorption decreases roughly monotonically with wavelength (approximately as 1/λ to 1/λ³) — a broadband absorber that penalizes shorter wavelengths most, and explains why pigmented skin requires higher fluence or longer wavelengths for equivalent penetration • Water: negligible absorption through the visible range, but rises sharply beyond ~900–1000nm and dominates completely beyond ~1300nm — this sets the practical upper edge of the therapeutic window
Between these three curves lies a trough — roughly 600–1200nm — where none of the three absorbers is strong. This is the therapeutic window exploited by every red/NIR-activated photosensitizer and by NIR photobiomodulation devices.
Moving a photosensitizer's activation wavelength from 630nm (Photofrin, 1st generation) to 664nm (many chlorins) to 689nm (some bacteriochlorins/newer agents) trades a modest drop in singlet-oxygen quantum yield for a substantial gain in tissue penetration — each ~30nm redshift can add roughly 1–2mm of usable treatment depth under this model.
A photosensitizer is only useful at wavelengths where it actually absorbs strongly (its Q-band), and that absorption must also fall inside the therapeutic window to reach depth. Photofrin's clinically-used Q-band sits at 630nm — inside the window, but at its shallow, hemoglobin-affected edge. Second- and third-generation photosensitizers (chlorins, bacteriochlorins, phthalocyanines) were deliberately redesigned with redshifted, stronger Q-bands at 660–700nm+ specifically to buy extra depth from the same delivered power, at the cost of more complex synthesis and sometimes lower singlet-oxygen quantum yield per absorbed photon.
Once light transport becomes diffusive, the fluence rate as a function of depth is well approximated by a diffusion-theory-modified Beer-Lambert relationship: an exponential decay governed by a single effective attenuation coefficient that folds absorption and scattering together.
The full radiative transport equation is too complex to solve analytically for arbitrary geometries, but under the diffusion approximation — valid a few transport mean free paths from the source, as established in Stage 2 — it reduces to a much simpler diffusion equation. For a broad, semi-infinite beam normally incident on a semi-infinite scattering-absorbing medium, the steady-state solution predicts fluence rate decaying exponentially with depth:
φ(z) = φ₀ · e^(−µeff·z)
where φ₀ is the surface fluence rate, z is depth, and µeff is the effective attenuation coefficient:
µeff = √(3·µa·(µa + µs′))
µa is the absorption coefficient (set mainly by hemoglobin, melanin, water and photosensitizer at the chosen wavelength — Stage 3), and µs′ is the reduced scattering coefficient (Stage 2, typically 5–20 cm⁻¹ in soft tissue). Because µs′ generally dominates µa in the visible/red range, µeff is driven primarily by how absorption interacts with a large scattering background — small changes in µa (i.e. in wavelength) can move µeff, and therefore penetration depth, substantially.
The "effective depth" (or penetration depth) δ = 1/µeff is defined as the depth at which fluence rate has fallen to 1/e (≈37%) of its surface value — it is a direct, single-number readout of how deep a given wavelength can usefully reach, and it is exactly the quantity this model plots as the tissue color gradient falls off with depth.
Because µeff decreases as wavelength increases through the therapeutic window, the same surface power produces a visibly flatter, slower-decaying exponential curve at 700nm than at 450nm. Doubling surface power (φ₀) shifts the entire curve upward by a constant factor but does not change its decay rate — it buys extra depth only logarithmically (each doubling of power adds ln(2)/µeff to the depth at which any fixed threshold is crossed). Changing wavelength changes µeff itself, which reshapes the whole curve — a fundamentally more powerful lever than simply turning up the source.
At some depth, fluence rate falls below the minimum needed to sustain a cytotoxic rate of singlet-oxygen generation — even with abundant photosensitizer and oxygen present. That crossing point defines the maximum tumor thickness treatable by a given wavelength/power combination, and it is the single biggest constraint on which cancers PDT can be used to treat.
The photodynamic effect requires three things to coincide in the same tissue voxel at the same moment: an activatable photosensitizer molecule, adequate molecular oxygen, and light above the cytotoxic fluence-rate threshold. Photosensitizer distribution and oxygen tension are biological variables that can be partly managed (drug-light interval, fractionated dosing), but the light term is fixed once wavelength and surface power are chosen and decays exponentially with depth as derived in Stage 4.
This is why surface-illuminated PDT is clinically restricted almost entirely to superficial or endoscopically-accessible lesions: cutaneous basal cell carcinoma and actinic keratosis (skin, directly accessible), Barrett's esophagus dysplasia (endoscopic light delivery to mucosal/submucosal depth), and non-muscle-invasive bladder carcinoma in situ (whole-bladder illumination via cystoscope). In each case the treatable lesion thickness — typically a few millimeters up to roughly a centimeter depending on wavelength and power — happens to match the natural thickness of the target tissue layer.
For bulkier tumors that exceed the surface-illumination depth limit, the clinical answer is not more surface power — logarithmic returns make that inefficient and often unsafe — but a different delivery geometry: interstitial fiber insertion, placing one or more cylindrical diffusing fibers directly inside the tumor mass so that no point in the target volume is more than a couple of effective-depths away from a light source.
Several complementary strategies extend PDT's clinical reach beyond the basic surface-illumination depth limit derived in this model:
• Longer-wavelength photosensitizers: shifting activation into the 700nm+ range (bacteriochlorins, some phthalocyanines) lowers µeff and directly increases both effective depth and maximum treatable depth, as shown by moving the wavelength slider toward the red end of this model • Fractionated / metronomic light delivery: splitting the same total dose into multiple lower-fluence-rate sessions (or a continuous low-rate infusion) can improve oxygen replenishment between photodynamic events, partially offsetting the hypoxia that high fluence rates induce and improving effective treatment depth for a given total dose • Interstitial fiber arrays with real-time dosimetry: multiple cylindrical diffusing fibers placed stereotactically through the tumor volume, often with fiber-optic fluence sensors providing real-time feedback, allow treatment of centimeter-scale bulky tumors (e.g. head & neck, prostate, pleural mesothelioma) that would be entirely inaccessible to surface illumination • Photosensitizer and delivery optimization: liposomal or nanoparticle carriers, and optimized drug-light intervals, can concentrate photosensitizer preferentially in tumor tissue, lowering the effective threshold power needed at any given depth