Sources and References

Primary texts — Szabo, T.L., Diagnostic Ultrasound Imaging: Inside Out, 2nd ed., Academic Press, 2014; Cobbold, R.S.C., Foundations of Biomedical Ultrasound, Oxford University Press, 2007.

Supplementary texts — Hill, C.R., Bamber, J.C., ter Haar, G.R. (eds.), Physical Principles of Medical Ultrasonics, 2nd ed., Wiley, 2004; Kinsler, L.E. et al., Fundamentals of Acoustics, 4th ed., Wiley, 2000; Prince, J.L. and Links, J.M., Medical Imaging Signals and Systems, 2nd ed., Pearson, 2014.

Online resources — MIT OpenCourseWare 6.555 “Biomedical Signal and Image Processing”; Stanford EE 369B lecture notes on medical imaging; IEEE UFFC open tutorials on transducer modeling; FDA Guidance on Marketing Clearance of Diagnostic Ultrasound Systems and Transducers; AIUM “Medical Ultrasound Safety” and output display standards (ODS); IEC 61157 and IEC 60601-2-37 publicly summarized.

---

Chapter 1: Acoustic Wave Propagation in Tissue

Diagnostic ultrasound exploits mechanical waves in the 1–50 MHz range to interrogate soft tissue. The same wave equation that governs a vibrating string or an organ pipe, when adapted to a compressible fluid-like medium, underlies every B-mode image, every Doppler spectrogram, and every therapeutic sonication.

1.1 The Linear Wave Equation

For small-amplitude disturbances, conservation of mass and Newton’s second law, combined with a linear equation of state \( p = \rho_0 c^2 s \) (where \( s \) is the condensation), yield the scalar wave equation

\[ \nabla^2 p - \frac{1}{c^2}\frac{\partial^2 p}{\partial t^2} = 0, \]

with sound speed \( c = \sqrt{K/\rho_0} \). In soft tissue, \( c \approx 1540 \) m/s, \( \rho_0 \approx 1050 \) kg/m\(^3\), giving a characteristic acoustic impedance \( Z = \rho_0 c \approx 1.62 \) MRayl.

1.2 Plane Waves and Intensity

A harmonic plane wave \( p(x,t) = P_0 \cos(\omega t - k x) \) carries time-averaged intensity

\[ I = \frac{P_0^2}{2 Z}. \]

Clinical diagnostic intensities (spatial-peak temporal-average, \( I_{SPTA} \)) are regulated below 720 mW/cm\(^2\) for most applications. Thermal and mechanical indices (TI, MI) displayed on scanners derive directly from these quantities.

1.3 Attenuation and Absorption

Tissue attenuation follows an approximately linear-with-frequency law:

\[ \alpha(f) = \alpha_0 f^y, \quad y \in [1, 1.3], \]

with \( \alpha_0 \approx 0.5 \) dB cm\(^{-1}\) MHz\(^{-1}\) in soft tissue. Attenuation couples image depth and resolution: higher frequency buys finer axial resolution but restricts penetration. A 3 MHz cardiac probe penetrates \( \sim 20 \) cm; a 15 MHz dermatology probe resolves \( \sim 0.1 \) mm but only to a few centimeters.

1.4 Scattering Regimes

Tissue parenchyma scatters ultrasound through sub-wavelength inhomogeneities. Rayleigh scattering (scatterer size \( \ll \lambda \)) produces the characteristic speckle texture whose statistics follow a Rayleigh distribution for fully developed speckle. Resolvable scatterers at scales comparable to \( \lambda \) produce specular echoes used for organ boundary delineation.

Chapter 2: Transducers and Beam Formation

2.1 The Piezoelectric Element

Lead zirconate titanate (PZT) and more recently single-crystal PMN-PT convert electrical energy to mechanical vibration via the piezoelectric effect. A plate of thickness \( t \) resonates at

\[ f_0 = \frac{c_p}{2 t}, \]

where \( c_p \) is the plate’s longitudinal sound speed. Matching layers of quarter-wavelength thickness and impedance \( Z_m = \sqrt{Z_{PZT} Z_{tissue}} \) maximize energy transfer into tissue. Backing layers broaden bandwidth at the cost of sensitivity, a classical Q-versus-sensitivity compromise.

2.2 Field from an Aperture

For an aperture of size \( D \) radiating at wavelength \( \lambda \), the near-field (Fresnel) extent is

\[ N = \frac{D^2}{4\lambda}, \]

beyond which the beam diverges with half-angle \( \sin\theta = 1.22\lambda/D \). Focusing lenses or phased-array delays place a focal zone near a desired depth; dynamic receive focusing refocuses electronically at every depth.

2.3 Phased and Linear Arrays

Modern scanners use 64–256 element linear, curvilinear, or phased arrays. By applying time delays \( \tau_n \) to element \( n \), the array steers and focuses a beam:

\[ \tau_n = \frac{r_f - \sqrt{(x_n - x_f)^2 + z_f^2}}{c}. \]

Apodization weighting controls side-lobe level. 2D matrix arrays (e.g., 3000+ elements) enable real-time 3D imaging, with micro-beamforming integrated into the probe handle to limit cable count.

2.4 Resolution Metrics

Axial resolution is set by pulse length, \( \Delta z \approx c \tau_p / 2 \), typically 1–2 wavelengths. Lateral resolution equals the beam width at focus, \( \Delta x \approx F\lambda/D \) (F-number times wavelength). Elevational resolution, governed by the out-of-plane aperture, is the weakest axis in 1D arrays and motivates 1.5D and 2D designs.

Chapter 3: Scanning Modes and Data Acquisition

3.1 A-Mode, B-Mode, and M-Mode

A-mode displays amplitude versus time (depth) for a single line; it survives in ophthalmic biometry. B-mode assembles many lines into a 2D grayscale image using log-compressed envelope detection. M-mode plots a single line versus time to resolve cardiac wall motion at millisecond scales.

3.2 Doppler Techniques

Blood flow produces frequency shifts \( f_d = 2 v \cos\theta f_0/c \). Continuous-wave Doppler measures high velocities without aliasing but lacks range resolution. Pulsed Doppler resolves range at the cost of a Nyquist limit \( v_{max} = c \cdot PRF / (4 f_0 \cos\theta) \). Color flow imaging estimates mean velocity on a grid using autocorrelation of quadrature-demodulated echoes (the Kasai estimator). Power Doppler ignores sign and magnitude, displaying the integrated signal power, giving superior sensitivity to slow flow.

3.3 Plane-Wave and Ultrafast Imaging

Conventional line-by-line scanning limits frame rates to \( \sim 50 \) Hz. Coherent plane-wave compounding transmits a single unfocused wave, beamforms in parallel, and compounds several angles. Frame rates of 5–20 kHz enable shear-wave elastography, ultrafast Doppler, and functional ultrasound neuroimaging.

3.4 Harmonic and Contrast Imaging

Tissue nonlinearity produces harmonics during propagation; tissue harmonic imaging filters the fundamental to improve contrast and suppress near-field clutter. Microbubble contrast agents (encapsulated perfluorocarbon gas, 1–5 µm) resonate strongly and are routinely imaged at the second harmonic or via pulse-inversion sequences that cancel linear echoes.

Chapter 4: Image Formation and Signal Processing

4.1 The Pulse-Echo Model

Received RF on channel \( n \) is modeled as a convolution of the transmit pulse with the tissue reflectivity function along the beam path, plus electronic noise. Delay-and-sum (DAS) beamforming aligns channel data:

\[ s(t) = \sum_{n=1}^{N} w_n \, r_n\!\left(t + \tau_n\right). \]

Adaptive beamformers (MVDR, coherence factor, DMAS) improve contrast and resolution at increased computational cost.

4.2 Envelope Detection and Log Compression

The analytic signal \( s_a = s + j \mathcal{H}\{s\} \) supplies the envelope \( |s_a| \). Log compression maps a 60–80 dB dynamic range into 8-bit display values, typically via \( D = a \log_{10}(|s_a|) + b \) with clinician-adjustable gain and dynamic range controls.

4.3 Speckle Statistics

For fully developed speckle, envelope amplitude follows a Rayleigh distribution, so speckle contrast \( \sigma/\mu \approx 0.52 \). Spatial compounding (averaging images from different transmit angles) and frequency compounding reduce speckle at some loss of resolution.

Chapter 5: Flow, Microscopy, and Elastography

5.1 Spectral Doppler Estimation

A gated range cell produces a slow-time signal whose short-time Fourier transform yields the spectral waveform. Clinical metrics include peak systolic velocity (PSV), end-diastolic velocity (EDV), resistive index \( RI = (PSV - EDV)/PSV \), and pulsatility index.

5.2 Super-Resolution Microscopy

Ultrasound localization microscopy (ULM) tracks individual microbubbles below the diffraction limit, producing microvascular maps with \( \sim 10 \) µm resolution deep in tissue—an ultrasound analog of PALM/STORM optical microscopy.

5.3 Elastography

Tissue stiffness correlates with pathology (fibrosis, tumors). Two principal techniques:

• Quasi-static (strain) elastography compares pre- and post-compression RF to estimate axial strain.
• Shear-wave elastography induces a transient shear wave by acoustic radiation force and tracks its propagation speed \( c_s = \sqrt{G/\rho} \), yielding the shear modulus \( G \).

Chapter 6: Therapy, Drug Delivery, and Safety

6.1 Bioeffects

Bioeffects partition into thermal and mechanical mechanisms. The bioheat equation

\[ \rho c_t \frac{\partial T}{\partial t} = k \nabla^2 T - w_b c_b (T - T_a) + Q, \]

with \( Q = 2\alpha I \) captures ultrasound-induced heating. Inertial cavitation, the violent collapse of bubble nuclei, dominates mechanical damage; the mechanical index \( MI = p_{neg}/\sqrt{f} \) summarizes risk.

6.2 High-Intensity Focused Ultrasound (HIFU)

HIFU focuses kilowatts of acoustic power into millimeter-scale volumes, ablating tumors via coagulative necrosis (thermal dose \( t_{43} \)). Clinical systems target uterine fibroids, prostate tumors, and, under MR guidance, brain targets through the intact skull using aberration-corrected arrays.

6.3 Targeted Drug and Gene Delivery

Microbubbles loaded with or co-administered with drugs enhance endothelial permeability upon sonoporation. Low-intensity focused ultrasound plus microbubbles transiently and reversibly opens the blood-brain barrier, a technique advancing through early-phase clinical trials for Alzheimer’s disease and glioblastoma.

6.4 Safety Standards and ALARA

Output display standards require scanners to show TI (thermal index) and MI in real time. The ALARA principle—as low as reasonably achievable—governs scanning practice: minimize exposure time, reduce output power to the lowest level that yields diagnostic quality, and avoid prolonged fetal or ocular exposure.