Balancing focal dose against overlying tissue safety as depth changes
Unlike light, which scatters and absorbs so strongly in tissue that it is confined to superficial applications, ultrasound attenuates gradually and can be dynamically focused at essentially any depth reachable by an acoustic window. That flexibility is exactly why dosimetry planning starts with a precise map of the acoustic path: the target depth and the layer-by-layer properties of everything the beam must cross to get there.
An ultrasound beam loses energy as it propagates through tissue via two combined mechanisms: absorption (acoustic energy converted to heat through viscous and relaxation losses) and scattering (energy redirected out of the beam path by small inhomogeneities). Together these are captured by the attenuation coefficient α, typically expressed in dB/cm/MHz because absorption in soft tissue rises roughly linearly with frequency.
The governing relation for pressure amplitude decay is exponential with depth:
p(z) = p₀ · e^(−α·z) (or, in decibels: Loss(dB) = α[dB/cm/MHz] × f[MHz] × z[cm])
For representative soft tissue, α ≈ 0.5–1 dB/cm/MHz. A 1 MHz beam traveling 10 cm therefore loses roughly 5–10 dB (a 3–10× drop in intensity) before it ever reaches the target — and that loss scales up directly with frequency, which is the central tension the rest of this planner explores.
Different tissues attenuate very differently: fat and muscle are moderate absorbers, lung and bone are extreme (bone can attenuate 10–20× more than soft tissue per unit length), and fluid-filled structures like bile or urine are nearly lossless. A path that crosses bone or aerated lung is often simply avoided in planning rather than compensated for.
Because attenuation is roughly proportional to frequency, doubling frequency to sharpen focal resolution roughly doubles the dB lost per centimeter of tissue traversed — depth and precision are always in tension, and every dosimetry plan is a negotiation between the two.
Each tissue layer the beam crosses — skin, subcutaneous fat, muscle, and the target organ — has its own characteristic acoustic impedance Z = ρc (density × speed of sound). At every interface between layers with different impedance, a fraction of the incident energy reflects rather than transmits, governed by the reflection coefficient:
R = [(Z₂ − Z₁) / (Z₂ + Z₁)]²
Soft-tissue-to-soft-tissue interfaces (e.g., fat/muscle) typically reflect only a few percent of incident energy, so most transmits forward. But large impedance mismatches — soft tissue to bone (Z ratio ~4–7×) or soft tissue to air (Z ratio >3000×) — reflect nearly all incident energy, which is why acoustic coupling gel and careful acoustic-window selection are mandatory, and why gas-filled bowel or lung in the beam path can make a target effectively unreachable.
Pre-treatment imaging (diagnostic B-mode ultrasound, CT for bone/gas mapping, or MRI for soft-tissue contrast) is used to trace the exact ray path from candidate transducer positions to the target, tag each layer's thickness and estimated α, and sum the path attenuation before any energy is delivered.
Every subsequent planning decision — transducer frequency, aperture size, electronic focusing scheme, and output power — is downstream of two numbers fixed at this stage: the target depth z and the cumulative path attenuation to reach it. A shallow target (2–3 cm, e.g., a superficial tumor or joint) tolerates high frequency because the attenuation penalty over a short path is small, buying a tight, precise focal spot. A deep target (10–15 cm, e.g., liver or uterine fibroids) forces a lower frequency to keep the path loss and required output power within achievable and safe limits, at the cost of a larger, less precise focal zone.
This depth-vs-frequency tradeoff, formalized quantitatively in later stages, is why ultrasound dosimetry planning is fundamentally a depth-dependent optimization problem rather than a fixed protocol.
Directly in front of a flat or weakly curved transducer face, the acoustic field is not yet a clean, converging beam. It is a turbulent interference pattern — the Fresnel (near) field — where waves radiating from every point on the aperture combine constructively in some places and destructively in others, producing intensity peaks and nulls that shift with distance rather than a smooth decline.
A real transducer face is not a single point source — it is an extended aperture, and Huygens' principle treats every point on that aperture as an independent wavelet source. Close to the face, the path-length differences between wavelets originating at the center versus the edge of the aperture are large relative to a wavelength, so the wavelets arrive out of phase at many points, producing a complex, non-monotonic interference pattern of on-axis and off-axis peaks and nulls.
The extent of this near field is given by:
N = a² / λ
where a is the aperture radius and λ = c/f is the wavelength in tissue. For a 1.2 cm radius transducer at 1 MHz (λ ≈ 1.54 mm ≈ 0.154 cm), N ≈ (1.2)²/0.154 ≈ 9.4 cm — meaning nearly a decimeter of tissue can lie within this unstable interference regime before the beam settles into predictable far-field behavior.
Near-field hot spots can locally exceed the average beam intensity by several decibels even though the nominal output power is unchanged — a real safety consideration for any overlying tissue that happens to sit at a constructive-interference peak rather than a null.
Because near-field intensity is not simply decaying with depth, a dosimetry plan cannot treat overlying tissue exposure as a smooth, predictable curve within this region — it must account for the possibility of local maxima. This is one reason therapeutic transducers are often designed as focused (concave) apertures or phased arrays rather than flat pistons: geometric or electronic focusing effectively begins converging the beam from the aperture itself, shrinking or eliminating the problematic near-field region and shifting the primary energy concentration directly to the intended focal depth.
The near-field length N also scales directly with frequency (since λ shrinks as f rises) and with the square of aperture radius — so higher-frequency, larger-aperture transducers push the unpredictable interference zone deeper into tissue, which must be factored into where the transducer is positioned relative to the skin surface and target.
Beyond the near-field/far-field transition distance N, the wavelets from across the transducer aperture fall into coherent alignment and the beam converges toward a well-defined focal zone. This is where a therapeutic ultrasound system deliberately concentrates its energy — the entire premise of focused ultrasound depends on a predictable, controllable focal geometry emerging beyond this transition.
In the far field of an unfocused aperture, the beam behaves like it originates from a point source, with intensity falling off approximately as 1/z² from simple spherical spreading (on top of tissue attenuation). Therapeutic transducers avoid wasting energy this way by shaping the aperture concave (geometric focusing) or by phasing an array of elements (electronic focusing) so that all wavelets arrive in phase specifically at the intended target depth — concentrating energy into a small, high-intensity focal volume rather than letting it spread.
The focal spot dimensions are diffraction-limited: lateral resolution scales with λ × f-number, and axial (depth-wise) resolution is typically several times larger than lateral resolution. A tighter f-number (larger aperture relative to focal depth) and higher frequency both shrink the focal spot — improving spatial precision, which matters when the target sits adjacent to critical structures.
Focal gain G describes the intensity multiplication achieved at the focus relative to the intensity that would exist without focusing at the same depth. For a spherically focused aperture, gain scales approximately with the ratio of aperture area to focal spot area — meaning a larger aperture, a shorter focal length, or both, produce a sharper, more intense focus.
This gain is precisely what allows focused ultrasound to reach ablative or bioeffect-triggering intensities (hundreds to thousands of W/cm² at the focus) while the transducer face itself, and the overlying tissue the beam passes through, are exposed to intensities that are one to two orders of magnitude lower. Focal gain is therefore the physical mechanism that makes deep, precise, non-invasive therapeutic dosing possible at all.
Single-element concave (bowl-shaped) transducers achieve geometric focusing at a fixed depth set by their physical curvature — simple and robust, but the focal depth cannot be changed without physically moving or swapping the transducer. Phased-array transducers instead apply small per-element time delays so that wavelets from every element arrive in phase at a chosen focal point — this electronic focusing (and electronic steering) lets the planner move the focus to different depths and lateral positions within milliseconds, without repositioning the probe, which is essential for treating extended target volumes or adjusting for patient motion in real time.
Tissue absorbs acoustic energy exponentially with path length, and that absorption grows with frequency. To deliver an adequate therapeutic dose at a deep target, the planner must therefore raise transducer output power as depth increases — all while continuously checking that the overlying tissue the beam passes through does not itself accumulate unsafe intensity or heating along the way.
To deliver a target focal intensity I_focal at depth z, the required source intensity I₀ at the transducer face must overcome the cumulative path loss:
I₀ = I_focal × 10^(α·f·z / 10) / G
where α is the attenuation coefficient (dB/cm/MHz), f is frequency (MHz), z is depth (cm), and G is the focal gain from Stage 3. Because the exponent grows linearly with both frequency and depth, the required source power grows exponentially with either — a target twice as deep, at the same frequency, does not need merely double the power; it needs power scaled by 10^(α·f·Δz/10), which compounds quickly.
This is the direct engineering consequence of Stage 1's physics: attenuation-compensated planning is the arithmetic of paying, in transducer output watts, for every decibel that tissue will absorb before the beam reaches the target — while never losing sight of what that same path loss does to the tissue doing the absorbing.
Because path loss scales with both α and frequency, doubling depth at a fixed frequency can require an order of magnitude more source power to maintain the same focal intensity — which is precisely why deep targets are treated at lower frequencies: the power budget and the overlying-tissue heating budget both become unworkable otherwise.
Lower frequency (0.25–1 MHz) penetrates deeper for a given power budget because attenuation is lower, but the longer wavelength produces a larger, less precise focal spot (diffraction-limited resolution scales with λ) — acceptable for deep, large-volume targets like liver or fibroid ablation, but too coarse for fine, precisely bounded lesions.
Higher frequency (1–3 MHz and above) produces a tighter, more precise focus, ideal for superficial or small targets (e.g., dermatologic or ophthalmic applications), but pays for that precision with steeply higher overlying-tissue attenuation and heating per centimeter of path, which sharply limits achievable depth and overlying-tissue safety margin.
Every therapeutic ultrasound platform frequency (typically 0.25–3 MHz for deep targets, rising into the tens of MHz for superficial or ophthalmic applications) represents this same negotiated tradeoff, chosen to match the specific target depth and required precision.
Raising source power to compensate for depth does not only raise focal intensity — it proportionally raises the intensity, and therefore the absorbed-power density and local heating, in every layer of overlying tissue the beam crosses on the way to the focus. Continuous-wave high-power exposure can produce clinically significant heating well before the beam even reaches the target.
Planners manage this with pulsed (gated) exposure: delivering energy in short bursts at a controlled duty cycle (commonly 10–50%) allows heat generated during "on" periods to diffuse away during "off" periods, keeping the time-averaged overlying-tissue temperature rise within safe bounds while the focal region — where the beam is concentrated by focal gain — still accumulates enough dose over the full treatment to be therapeutically effective.
The final step of planning confirms, quantitatively, that the compensated power delivery designed in Stage 4 lands within safe therapeutic ranges everywhere it matters: adequate focal intensity to achieve the intended bioeffect, and mechanical index (MI) and thermal index (TI) at or below regulatory ceilings in both the focal zone and every overlying tissue layer.
The Mechanical Index (MI) estimates the risk of non-thermal, mechanical bioeffects — chiefly cavitation, the violent growth and collapse of gas bubbles under negative acoustic pressure — and is defined as:
MI = p_r⁻ / √f
where p_r⁻ is the peak rarefactional (negative) pressure in MPa, derated for tissue attenuation, and f is frequency in MHz. Dividing by √f reflects that cavitation risk falls as frequency rises for a given peak pressure. The FDA regulatory ceiling for MI in general diagnostic use is 1.9.
The Thermal Index (TI) estimates the risk of thermal bioeffects by comparing acoustic power output to the power needed to raise tissue temperature by 1°C under standardized conditions, with separate soft-tissue (TIS), bone (TIB), and cranial (TIC) variants reflecting how differently those tissues absorb and dissipate heat. Both indices are calculated in real time by the system and displayed to the operator under the FDA's Output Display Standard (ODS) — the regulatory mechanism that makes MI/TI compliance an explicit, continuously visible part of every scan or treatment.
MI and TI are not merely diagnostic-imaging concepts — therapeutic focused ultrasound planning uses the same framework, deliberately operating at a controlled, elevated index specifically at the focus (where the bioeffect is intended) while engineering the overlying path, via frequency selection, focal gain, and duty cycling, to stay under the same ceilings everywhere else.
Final plan verification walks the entire beam path from transducer face to target and checks three things at every depth increment: (1) is cumulative intensity, and therefore MI/TI, below the safety ceiling in every overlying layer; (2) does focal intensity at the target depth meet or exceed the minimum required for the intended therapeutic bioeffect (thermal ablation, sonodynamic activation, or mechanical disruption, depending on application); and (3) does the transition between "safe overlying path" and "adequate focal dose" happen sharply enough, via focal gain, that no intermediate tissue receives a dose that is both unnecessary and unsafe.
When verification fails — commonly because a deep target at high frequency would require overlying intensity beyond safe limits to reach adequate focal dose — the plan is revised: lower frequency to reduce path loss, increase aperture or reduce f-number to raise focal gain, reduce duty cycle to manage heating, or in some cases reposition the acoustic window to shorten or improve the tissue path.
Diagnostic and therapeutic ultrasound safety in the United States is governed primarily by the FDA's Output Display Standard, which mandates real-time MI and TI display, alongside guidance documents from the American Institute of Ultrasound in Medicine (AIUM) and the "As Low As Reasonably Achievable" (ALARA) principle for diagnostic exposure. Therapeutic and ablative focused-ultrasound systems (e.g., MR-guided HIFU for uterine fibroids or essential tremor) undergo their own device-specific FDA clearance, in which the intentionally elevated focal-zone MI/TI is justified against the therapeutic benefit while overlying-path exposure is held to standard safety ceilings — the same balancing act this entire dosimetry planning workflow is designed to verify before a single joule reaches the patient.