Acoustic Computation: From Effective Medium Theory to Biomedical Ultrasound Imaging

Theoretical Foundations of Effective Medium Theory and Homogenization

• Definition and Core Principles:

  • Effective Medium Theory (EMT) provides analytical and semi-analytical methods to simplify complex composite materials into homogeneous media characterized by equivalent effective parameters: mass density $\rho_e$ and bulk modulus $\neq_e$.
  • EMT significantly reduces computational complexity compared to grid-based numerical methods (such as Finite Element Methods [FEM] and Boundary Element Methods [BEM]) while accurately preserving long-wavelength acoustic responses and material resonance behavior.
  • EMT acts as a fast response function or a physics-informed model that enables the reverse design of high-performance metamaterials.

• Complementarity Between Homogenization and Imaging:

  • The homogenization process and the imaging process are physically complementary, established through the relationship between spatial coordinates and time-frequency transformations.
  • Homogenization Process: Encodes spatial structural information into the frequency domain. Replacing discrete impedance barriers with an equivalent uniform material manifests as frequency dispersion or local resonance in effective parameters.
  • Imaging Process (e.g., A-mode imaging): Inverts time-domain echo information (time-of-flight) back into spatial position coordinates to locate internal material interfaces.
  • Time and frequency domains are interconverted via the Fourier transform or its inverse version.

• Static Homogenization (Long-Wavelength Limit $L \ll \lambda$):

  • Under static or long-wavelength limits, the effective properties perceived by sound waves reflect volume-averaged responses analogous to series and parallel electrical resistor networks.
  • Wood's Formula for Bulk Modulus: For a two-phase fluid-solid composite where Phase 1 forms a topologically connected fluid matrix:         $\frac{1}{\kappa_e} = \frac{f_1}{\kappa_1} + \frac{f_2}{\kappa_2}$         where $f_1 + f_2 = 1$ represent volume fractions. If Phase 1 is solid, Poisson's ratio $\nu_1$ must be introduced into the model.
  • Static Mass Density (Series vs. Parallel):
    • Series arrangement (layered structures perpendicular to propagation): $\rho_e = f_1 \rho_1 + f_2 \rho_2$.
    • Parallel arrangement (structures allowing relative motion between phases, e.g., hyperlenses): $\rho_e^{-1} = f_1 \rho_1^{-1} + f_2 \rho_2^{-1}$.
  • Berryman's Generalized Density Formula: Derived via first-order approximations in multiple scattering theory for a $d$-dimensional solid-fluid composite:         $\frac{\rho_e - \rho_1}{(d - 1)\rho_e + \rho_1} = f_2 \frac{\rho_2 - \rho_1}{(d - 1)\rho_2 + \rho_1}$
  • Limitations: At high volume fractions of solid inclusions, near-field interactions via evanescent waves and higher-order scattering channels require correction factors (similar to added mass effects in acoustic perforated plates).

• Dynamic Homogenization and Locally Resonant Metamaterials:

  • Deep-subwavelength monopole ($L \ll \lambda$) and dipole resonances shatter classical static limits, inducing anomalous effective properties including negative dynamic density and negative bulk modulus.
  • Dipole Resonator Effective Density (Lorentzian Form):$\rho_e = \rho_0 + \rho_m \frac{\omega^2}{\omega_d^2 - \omega^2}$         where $\omega_d$ is the dipole resonance frequency.
  • Monopole Resonator Effective Modulus:$\kappa_e^{-1} = \kappa_0^{-1} \left(1 - \alpha_m \frac{\omega^2}{\omega_m^2 - \omega^2}\right)$         where $\omega_m$ is the monopole resonance frequency.

• Extensions of EMT for Complex Biological Media:

  • Non-Hermiticity: Accounts for broken time-reversal symmetry in dissipative or gain media. Dissipation weakens propagating ultrasound signals, lowering the signal-to-noise ratio (SNR) in deep tissue. Broadband frequency attenuation and phase dispersion are governed strictly by the Kramers-Kronig relations:         $\text{Phase Velocity } c(\omega) \text{ and Attenuation } \alpha(\omega) \text{ linkage}$
  • Non-uniformity: Gradient index/impedance materials ($\rho(x)$, $\kappa(x)$) provide broadband impedance matching, focusing, and enhanced wave transmission through dense obstacles (such as cranial bones).

Wave Physics in Computational and Optical Inspired Imaging

• Optical Coherence Theory and Wave Interference:

  • Monochromatic scalar wave propagation through an aperture wavefront $S$ follows the Huygens-Fresnel integral:         $U(\mathbf{r}, \omega) = -\frac{i}{\lambda} \int_S U^{(0)}(\mathbf{r}', \omega) \frac{e^{i k s}}{s} \, dS$         where $k = \frac{\omega}{c} = \frac{2\pi}{\lambda}$ is the wavenumber and $s = |\mathbf{r}' - \mathbf{r}|$.
  • Total field under scatterer interaction:         $U(\mathbf{r}, \omega) = U^{(i)}(\mathbf{r}, \omega) + U^{(s)}(\mathbf{r}, \omega)$
  • First-Order Born Approximation for weak scattering in volume $V$:         $U(\mathbf{r}, \omega) = U^{(i)}(\mathbf{r}, \omega) + \int_V F(\mathbf{r}', \omega) U^{(i)}(\mathbf{r}', \omega) G(\mathbf{r} - \mathbf{r}', \omega) \, d^3\mathbf{r}'$         where scattering potential $F(\mathbf{r}, \omega) = \frac{k^2}{4\pi} [n^2(\mathbf{r}, \omega) - 1]$, $n(\mathbf{r}, \omega)$ is the refractive index, and free-space Green's function $G(\mathbf{r} - \mathbf{r}', \omega) = \frac{e^{i k |\mathbf{r} - \mathbf{r}'|}}{|\mathbf{r} - \mathbf{r}'|}$.
  • Zero-intensity interference points in multi-pinhole patterns correspond directly to phase singularities and phase dislocations.
  • Spatial coherence of backscattered acoustic waves determines tissue speckle statistics, described mathematically by the van Cittert-Zernike theorem.

• Transport of Intensity Equation (TIE):

  • TIE establishes a non-iterative, deterministic relationship between the axial derivative of wave intensity and the transverse phase gradient:         $-k \frac{\partial}{\partial z} I(\mathbf{x}, z) = \nabla_\perp \cdot (I(\mathbf{x}, z) \nabla_\perp \phi(\mathbf{x}, z))$         where $\nabla_\perp$ is the transverse gradient operator.
  • Axial derivative approximation via finite intensity differences across defocus distance $\Delta z$:         $\frac{\partial}{\partial z} I(\mathbf{x}, z) \approx \frac{I(\mathbf{x}, z + \Delta z) - I(\mathbf{x}, z - \Delta z)}{2\Delta z}$
  • Offers a non-iterative pathway for ultrasound quantitative phase recovery across depth-defocused planes.

• Ptychographic Phase Retrieval:

  • Requires translational invariance with spatial overlap between adjacent illumination beam positions (shifts $\mathbf{R}$).
  • Complex exit wave function:         $\psi(\mathbf{r}, \mathbf{R}) = O(\mathbf{r}) P(\mathbf{r} - \mathbf{R})$         where $O(\mathbf{r})$ is the 2D sample transmission function and $P(\mathbf{r})$ is the illumination function.
  • Iterative object update equation:         $O_{g,n+1}(\mathbf{r}) = O_{g,n}(\mathbf{r}) + \frac{|P(\mathbf{r} - \mathbf{R})|}{|P_{\text{max}}(\mathbf{r} - \mathbf{R})|} \frac{P^*(\mathbf{r} - \mathbf{R}) E_{c,g,n}(\mathbf{r}, \mathbf{R})}{|P(\mathbf{r} - \mathbf{R})|^2 + \alpha}$         where $E_{c,g,n}(\mathbf{r}, \mathbf{R}) = \beta (\psi_{c,n}(\mathbf{r}, \mathbf{R}) - \psi_{g,n}(\mathbf{r}, \mathbf{R}))$, $\alpha$ is a stabilization parameter, and $\beta$ is the update step size.
  • Sum Squared Error (SSE) metric over $N$ array elements:         $\text{SSE} = \frac{(|\Psi(\mathbf{k}, \mathbf{R})|^2 - |\Psi_{g,n}(\mathbf{k}, \mathbf{R})|^2)^2}{N}$

• Comparative Computational Complexity Across Imaging Modalities:

  • Conventional Focused Imaging ($128$ focused beams, $4$ focal depths): Sequential transmissions and point-by-point delay-and-sum (DAS). Complexity scales as $\mathcal{O}(N_e \times N_{\text{px}})$, yielding low frame rates ($\sim 25\,\text{fps}$).
  • Single-Angle Plane-Wave Imaging ($1$ transmission): Single DAS operation, complexity $\mathcal{O}(N_e \times N_{\text{px}})$, achieves ultrahigh frame rates ($\sim 18{,}000\,\text{fps}$).
  • Plane-Wave Compounding ($17$ angles): Coherently sums $17$ plane-wave acquisitions. Complexity scales linearly as $\mathcal{O}(17 \times N_e \times N_{\text{px}})$, achieving $\sim 1000\,\text{fps}$.
  • Plane-Wave Compounding ($40$ angles): Complexity $\mathcal{O}(40 \times N_e \times N_{\text{px}})$, frame rate $\sim 350\,\text{fps}$.
  • Iterative Ptychography: Dominated by Fast Fourier Transforms (FFTs) scaling as $\mathcal{O}(N_{\text{px}} \log N_{\text{px}})$ per iteration.

• Ultrasound Governing Equations and Photoacoustics:

  • Constant-density acoustic Helmholtz equation:         $\nabla^2 p(\mathbf{r}) + k^2(\mathbf{r}) p(\mathbf{r}) = 0$
  • Eikonal equation under high-frequency limit ($\lambda_0 \to 0$):         $|\nabla \tau(\mathbf{r}, c)|^2 = n^2(\mathbf{r})$         where $n(\mathbf{r}) = \frac{c_0}{c(\mathbf{r})}$ is refractive index and $\tau(\mathbf{r})$ is wavefront travel time.
  • Ultrasound Born approximation formulation:         $p(\mathbf{r}, \omega) = p^{(i)}(\mathbf{r}, \omega) + \int_V O(\mathbf{r}', \omega) p^{(i)}(\mathbf{r}', \omega) G(\mathbf{r} - \mathbf{r}', \omega) \, d^3\mathbf{r}'$         where scattering potential $O(\mathbf{k}, \mathbf{r}) = k^2 - k_0^2$ and 2D Green's function $G(\mathbf{r} - \mathbf{r}', \omega) = \frac{j}{4} H_0^{(1)}(k_0 |\mathbf{r} - \mathbf{r}'|)$.
  • Photoacoustic transient initial pressure rise under thermal confinement ($\tau_p \ll \tau_{\text{th}}$) and stress confinement ($\tau_p \ll \tau_s$):         $p_0(z, t) = \Gamma \mu_a F_0 e^{-\mu_0 z}$         where $\Gamma$ is the Grüneisen coefficient, $\mu_a$ is optical absorption, $F_0$ is incident laser fluence, and $\mu_0$ is acoustic attenuation.

Medical Ultrasound Computed Tomography (USCT)

• Rayleigh Wave Integrals and Single-Emitter Systems:
  • Velocity potential $\phi(x, y, z)$ from a radiating surface $S$ via Rayleigh integral:         $\phi(x, y, z) = \int_S \frac{v_0 e^{j(\omega t - k r)}}{2\pi r} \, dS$
  • Pressure field generated by a circular plane piston in a rigid baffle:         $p = \frac{i \omega \rho_0}{2\pi} e^{i \omega t} \int_S \frac{v e^{-i k R}}{R} \, dS$
  • A-scan Signal Classification: Transmission signals represent the fastest direct path from emitter to receiver; reflection signals represent backscattered/reflected paths.

• X-Ray Analogy and Hounsfield Units (HU):

  • X-ray attenuation values map to the Hounsfield Unit scale:         $\text{HU}_{\text{material}} = 1000 \times \frac{\mu_{\text{material}} - \mu_{\text{water}}}{\mu_{\text{water}}}$         yielding $0\,\text{HU}$ for distilled water and $-1000\,\text{HU}$ for air.

• USCT Reconstruction Algorithms:

  • Straight-line time-of-flight (TOF) integral for refractive index $n = \frac{v_0}{v}$:         $\int_{r_{kl}} n \, ds = v_0 T_{kl}$
  • Complex refractive index incorporating spatial absorption $\mu(\mathbf{x})$:         $\eta(\mathbf{x}) \approx \frac{c_0}{c(\mathbf{x})} + i \frac{\mu(\mathbf{x})}{k_0}$
  • Scattered field equation under first-order Born approximation:         $\hat{p}_{\text{sct}}(\mathbf{x}) = \omega^2 \int_{\mathbf{x}' \in D} \hat{G}(\mathbf{x} - \mathbf{x}') \hat{p}_{\text{inc}}(\mathbf{x}') \chi(\mathbf{x}') \, dA(\mathbf{x}')$
  • Distorted Born Iterative Method (DBIM) linear inverse solver step:         $p(\mathbf{r}, k) - p(\mathbf{r}, k_b) \approx \int \, d\mathbf{r}' [k^2(\mathbf{r}') - k_b^2(\mathbf{r}')] p(\mathbf{r}', k_b) G_b(\mathbf{r}, \mathbf{r}')$
  • Relative Residual Error (RRE) iteration criterion:         $\text{RRE} = \frac{\|p(\vec{\rho}, k) - p(\vec{\rho}, k_b)\|}{\|p(\vec{\rho}, k)\|}$

• USCT Experimental Parameters and Implementation:

  • Water tank temperature fixed at $18.1\,^\circ\text{C}$.
  • Transducers: Unfocused plane piston radiation sources operating at $1\,\text{MHz}$ center frequency ($1$-cycle burst), characteristic acoustic size $k a = 6.65$.
  • Target phantom: Soft rubber balloon film ($c = 1540\,\text{m/s}$, thickness $0.23\,\text{mm}$, outer diameter $7.3\,\text{mm}$) filled with saline solution ($c = 1718\,\text{m/s}$).
  • Distances: Fixed transmitter-to-balloon center distance = $104\,\text{mm}$, balloon-to-receiver distance = $91\,\text{mm}$. L-shaped arm rotated from $-60^\circ$ to $60^\circ$ in $1^\circ$ steps.

Dynamic Beamforming and Pulse-Echo Reconstruction Methods

• Delay-and-Sum (DAS) Mathematical Formulation:

  • Two-way round-trip Time of Flight (TOF) to scatterer $X_s = (x_s, z_s)$:         $\tau_i(X_s) = \frac{d_{\text{TX}}(X_s) + d_{\text{RX}}(X_s, x_i)}{c} - t_0$
  • Diverging wave transmission distance $d_{\text{TX}}$$d_{\text{TX}}(X_s) = \sqrt{(x_s - x_0)^2 + (z_s - z_0)^2} - \sqrt{H(|x_0| - L/2)(|x_0| - L/2)^2 + z_0^2}$         where virtual source coordinates $(x_0, z_0)$ for tilt angle $\theta$ and beam angle $\beta$ are:         $x_0 = \frac{L \sin(2\theta)}{2 \sin(\beta)}, \quad z_0 = -\frac{L \cos(\beta)}{2} + \frac{\cos(2\theta)}{\sin(\beta)}$
  • Plane-wave limit ($\beta \to 0^+$):         $\lim_{\beta \to 0^+} d_{\text{TX}}(X_s) = (\text{sgn}(\theta) L / 2 - x_s) \sin(\theta) + z_s \cos(\theta)$
  • Reception distance: $d_{\text{RX}}(X_s, x_i) = \sqrt{(x_i - x_s)^2 + z_s^2}$.
  • DAS Beamformed Radio-Frequency (RF) signal:         $s_{\text{bf}}(X_s) = \sum_{i \in [1, N_e]} s_i(\tau_i(X_s))$
  • In-Phase/Quadrature (I/Q) demodulated DAS beamforming:         $\text{IQ}_{\text{bf}}(X_s) = \sum_{i \in [1, N_e]} \text{IQ}_i(\tau_i(X_s)) e^{2j\pi f_c \tau_i(X_s)}$         constrained by f-number: $\frac{z_s}{2 |x_s - x_i|} \ge f_{\#}$.
  • DAS matrix sparsity ($q$-point interpolation):         $\text{sparsity}(M_{\text{DAS}}) \ge 1 - \frac{q}{n_s}$         GPU Sparse Matrix-Vector Multiplication (SpMV) acceleration yields $60\text{--}100\times$ speedup over CPU implementations using $64\times$ fewer nodes ($2$ GPUs vs $128$ CPU nodes).

• Motion-Compounded (MoCo) Imaging:

  • Phase delay angle extracted from inter-frame autocorrelation across $M$ compounding transmissions:         $\phi_{\text{MoCo}}(\theta, r) = \frac{1}{2} \angle \{M_1 M_2\}$
  • Coherent compound complex envelope:         $\tilde{s}_C(\theta, r) = \sum_{m=1}^M \tilde{s}_m \left(\theta, r + \left(m - \frac{M}{2}\right) \phi_{\text{MoCo}} \frac{4\pi c}{f_0}\right) e^{i m \phi_{\text{MoCo}}}$

• Null Subtraction Imaging (NSI):

  • Nonlinear beamforming method providing a $6\times$ improvement in spatial resolution over standard DAS while increasing computational load by only $40\%$.

• Tissue Speed of Sound Values and Effective Media:
  • Fat tissue: $1435\,\text{m/s}$
  • Muscle tissue: $1595\,\text{m/s}$
  • Blood vessels: $1616\,\text{m/s}$
  • Blood: $1580\,\text{m/s}$
  • Bone: $2500\,\text{m/s}$
  • Liver: $1567\,\text{m/s}$
  • 3D-Printed Phantom Materials: Agilus30 ($2035\,\text{m/s}$), Formlabs resin ($2591\,\text{m/s}$), VeroClear ($2473\,\text{m/s}$).
  • Kuster-Toksoz EMT Model for Red Blood Cells (concentration $\phi_i$):$\rho_{\text{ag}} = \phi_i \rho_1 + (1 - \phi_i) \rho_2, \quad \frac{1}{\kappa_{\text{ag}}} = \frac{\phi_i}{\kappa_1} + \frac{1 - \phi_i}{\kappa_2}$

Wave Propagation Through Aberrating Layers and Matrix Imaging

• Van Cittert-Zernike Theorem in Random Scattering Media:

  • Spatial covariance of backscattered intensity for diffuse targets at aperture spacing $\Delta \mathbf{X}_a = [x_a, y_a]^\top$:         $R(\Delta \mathbf{X}_a, z, f) = \int_{-\infty}^{+\infty} |P(\mathbf{X}_s, z, f) \chi(\mathbf{X}_s, z, f)|^2 e^{-2\pi j (\mathbf{X}'_s \Delta \mathbf{X}_a) / \lambda z} \, d\mathbf{X}_s$
  • Signal-to-Noise Ratio (SNR):         $\text{SNR} = A_R A_T \frac{\int [R_P(x)]^2 R_{\text{OR}}(x) \, dx}{\left( \int [R_P(x)]^2 \, dx \int R_{\text{OR}}^2(x) \, dx \right)^{1/2}}$         where $A_R$, $A_T$ are receiver/transmitter aperture areas, $R_{\text{OR}}$ is receiver autocorrelation, and $R_P(x)$ is spatial covariance.

• Time Reversal Mirror (TRM) and Ultrasound Matrix Imaging (UMI):

  • Derode et al. multiple-scattering experiment: $2000$ steel rods in water demonstrated acoustic time reversibility, achieving a spatial focus of $1.05\,\text{mm}$ (surpassing the classical aperture limit $1.2 \lambda z / a = 6.38\,\text{mm}$ by a factor of six).
  • UMI Matrix Operations:
    • Reflection matrix: $R_{kk}(z) = T(z) \times \Gamma(z) \times T^\top(z)$, connecting input/output focus modes through true transmission matrix $T(z)$ and target scattering matrix $\Gamma(z)$.
    • Distortion matrix $D(\mathbf{u}, \mathbf{r})$: Isolates wavefront aberrations relative to a homogeneous reference medium.
  • Spatial frequency spectrum from reflection matrix coefficients:         $|R(k_{\text{out}}, k_{\text{in}}, z)|^2 = |\tilde{\gamma}(k_{\text{out}} + k_{\text{in}}, z)|^2$
  • Filtered matrix removing multiple reflections/reverberations:         $R'_{xx}(z) = T_0^\dagger(z) \times R'_{kk}(z) \times T_0^*(z)$

• Transcranial Ultrasound Focusing Techniques:
  • Sound speed contrast: Skull bone $c_{\text{skull}} \approx 3000\,\text{m/s}$, brain tissue $c_{\text{brain}} \approx 1400\,\text{m/s}$.
  • Full-Waveform Inversion (FWI): 3D brain wave speed mapping requires $32\,\text{hours}$ on a $128$-node ($24$ CPU cores/node) cluster; GPU servers ($>1\,\text{PFLOPS}$) reduce computation to $<10\,\text{minutes}$.
  • Complementary Metamaterial (CMM) Tensors for cancelling aberrating layers:         $[\rho^{(c)}]^{-1} = \frac{\mathbf{A} [\rho^{(a)}]^{-1} \mathbf{A}^\top}{\det \mathbf{A}}, \quad \beta^{(c)} = \frac{\beta^{(a)}}{\det \mathbf{A}}$         where $\mathbf{A}$ is the coordinate transformation Jacobian matrix.
  • Phase-only holograms coupled with tungsten metalenses compensate for near-field phase distortion, enabling targeted bilateral blood-brain barrier (BBB) opening and transcranial drug delivery.

Electro-Acoustic Transducer Network Modeling and Ultrafast Compounding

• KLM Electro-Acoustic Circuit Model:

  • Lumped-parameter electrical circuit coupled with transmission lines representing mechanical structure layers (piezoelectric element, backing layer, matching layers).
  • Input electrical impedance $Z_{\text{in}}$ at electrical port:         $Z_{\text{in}} = \frac{1}{j \omega C_0} + j X_1 + \frac{Z_a}{\phi^2}$
  • Capacitance $C_0$ and inductance $X_1$:         $C_0 = \frac{\varepsilon A}{d}, \quad X_1 = \frac{h^2}{\omega^2 Z_0} \sin\left(\frac{\omega d}{c}\right)$
  • Piezoelectric crystal acoustic impedance $Z_0$ and electromechanical transformer ratio $\phi$:         $Z_0 = \rho c A, \quad \phi = \frac{\omega Z_0}{2 h \csc\left(\frac{\omega d}{2c}\right)}$         where $\rho$ is density, $c$ is speed of sound, $A$ is element area, $d$ is thickness, $\varepsilon$ is permittivity, and $h$ is the piezoelectric constant.
  • Equivalent acoustic load impedance $Z_a = \frac{Z_{L1} Z_{L2}}{Z_{L1} + Z_{L2}}$, with backing layer load $Z_{L1}$ and matching layer load $Z_{L2}$:         $Z_{L1} = Z_0 \frac{Z_1 + j Z_0 \tan(\omega d / 2c)}{Z_0 + j Z_1 \tan(\omega d / 2c)}, \quad Z_{L2} = Z_0 \frac{Z_2 + j Z_0 \tan(\omega d / 2c)}{Z_0 + j Z_2 \tan(\omega d / 2c)}$
  • Cardiac Probe Parameters: Piezoelectric crystal $Z_0 = 34.69\,\text{MRayl}$, backing layer $Z_1 = 7.8\,\text{MRayl}$, three matching layers ($8.5\,\text{MRayl}$, $6\,\text{MRayl}$, $3\,\text{MRayl}$), tissue medium ($1.5\text{--}2\,\text{MRayl}$), surface area $448\,\text{mm}^2$, $\ge 64$ array elements.

• Shear Wave Elasticity and Ultrafast Compounding:

  • Shear wave velocity $c_s = \sqrt{\frac{\mu}{\rho}} = \sqrt{\frac{E}{3\rho}}$ ($1\text{--}10\,\text{m/s}$ in soft tissue).
  • Compressional wave velocity $c_p = \sqrt{\frac{K}{\rho}}$ ($\sim 1540\,\text{m/s}$ in soft tissue).
  • A supersonic acoustic pushing pulse (speed $\sim 6\,\text{m/s}$) exceeding shear wave speed ($1\,\text{m/s}$) creates a Mach cone.
  • Frame rate metrics: Conventional focused scanning achieves $\sim 25\text{--}40\,\text{fps}$; single plane wave firing achieves up to $18{,}000\,\text{fps}$; $17$-angle plane-wave compounding yields $1000\,\text{fps}$.

Acoustic Computer Holography, Compressed Sensing, and Information Metrics

• Bessel Beams and Non-Diffracting Waves:

  • Zero-order and high-order Bessel beams resist diffraction and display self-healing properties.
  • Pressure distribution of a high-order Bessel beam carrying topological charge $M$:         $p_0(r, \theta) = \frac{1}{\max(r, r_0)} e^{i k r} e^{i M \theta}$

• Optical to Acoustic Holography Principles:

  • Historical Milestones: Gabor ($1948$ inline), Leith & Upatnieks ($1964$ off-axis), Goodman ($1967$ digital), Lohmann & Paris ($1966$ computer binary Fraunhofer).
  • Diffraction field pressure $p(\mathbf{X}_s, z, f)$ produced by 2D aperture amplitude distribution $A(\mathbf{X}_a, f)$:         $p(\mathbf{X}_s, z, f) = \frac{e^{j k z} e^{j k (\mathbf{X}'_s \mathbf{X}_s) / 2z}}{j \lambda z} \int_{-\infty}^{+\infty} \left[ A(\mathbf{X}_a, f) e^{j k \mathbf{X}'_a \mathbf{X}_a / 2z} \right] e^{-2\pi j \mathbf{X}'_s \mathbf{X}_a / \lambda z} \, d\mathbf{X}_a$
  • Iterative Angular Spectrum Algorithm (IASA):
    1. Compute 2D spatial Fourier transform at image plane: $P(k_x, k_y, z) = \iint_{-\infty}^{+\infty} p(x, y, z) e^{-j(k_x x + k_y y)} \, dx dy$
    2. Backpropagate spectrum to transducer plane using propagator $H(k_x, k_y, z) = e^{j z \sqrt{k^2 - k_x^2 - k_y^2}}$: $P(k_x, k_y, 0) = P(k_x, k_y, z) H(k_x, k_y, -z)$
    3. Forward propagate and evaluate real-space field via inverse Fourier transform: $p(x, y, z) = \frac{1}{4\pi^2} \iint_{-\infty}^{+\infty} P(k_x, k_y, z) \, dk_x dk_y$
    4. Iterate until Structural Similarity Index (SSIM) or Root Mean Square Error (RMSE) meets target threshold.

• Single-Sensor 3D Volumetric Compressed Sensing:

  • Linear forward system model: $\mathbf{u} = \mathbf{H} \mathbf{v} + \mathbf{n}$, where $\mathbf{u}$ is an $M$ -dimensional measurement vector, $\mathbf{H}$ is an $M \times N$ calibration/prediction matrix ($M < N$), $\mathbf{v}$ is the $N$-dimensional 3D spatial target vector, and $\mathbf{n}$ is noise.
  • Utilizes a single transducer paired with a rotating phase-coded aperture mask ($50$ rotations).
  • Forward pressure field for virtual sources $\mathcal{S}$ and Gaussian pulse $h(t) = \exp\left(-\frac{t^2}{2l^2}\right)$:         $p(t, 0) = |\mathcal{S}|^{-1} \sum_{j=0}^{|\mathcal{S}|-1} h(t - \tau(0, \mathbf{s}_j))$         where travel time c \tau(\theta, \mathbf{s}_j) = \frac{1}{c_1} [\mathbf{s}_j]_z + \frac{1}{c_0} \|\mathbf{s}_j - \theta\|^2$.\n * Echo signal autoconvolution: a(t, 0) = p(t, 0) * p(t, 0).\n * Fisher Information Matrix entry (i, j) for Cramér-Rao Lower Bound (CRLB) estimation:\n        I(\theta)_{i,j} = \frac{1}{\sigma^2} \frac{\partial a(\theta)^\top}{\partial [\theta]_i} \frac{\partial a(\theta)}{\partial [\theta]_j}$$

Systematic Framework, Emerging Paradigms, and Future Directions

• Integrated Acoustic Computation Roadmap:
  • Acoustic computation bridges Effective Medium Theory (homogenization, static/dynamic models, non-Hermiticity, non-uniformity) with Biomedical Ultrasound Imaging (USCT, DAS beamforming, aberration correction, quantitative ultrasound).
  • Optical inspirations (coherence theory, TIE, ptychography, holography) cross-link directly into emerging acoustic modalities (UMI, compressed sensing, acoustic holography, metamaterial lenses).

• Emerging Technological Frontiers:
  • AI and Transfer Learning: Adapting large, high-resolution CT/MRI clinical datasets via Convolutional Neural Networks (CNNs) and transfer learning to train acoustic segmentation models from small experimental ultrasound datasets.
  • All-Optical Ultrasound (OpUS): Replaces electronic piezoelectric elements with optical fiber generation (composite photothermal/thermal expansion thin films) and reception (Fabry-Pérot interferometers, Fiber Bragg Gratings).
  • Photonic Integrated Circuits (PIC): Enables ultrahigh-speed optical computation and interference processing directly on-chip, bypassing Peripheral Component Interconnect Express (PCIe) bus throughput bottlenecks to enable real-time 3D volumetric ultrasound reconstruction.