Ultrasound thermometry based on the temperature-dependent change in backscattered energy (CBE) of nonlinearly generated acoustic harmonics can localize heated regions and, after calibration, estimate tissue temperature. The method was demonstrated in homogeneous phantoms and ex vivo muscle at 13 MHz transmit with needle heating (Maraghechi et al., 2019), and follow-up work by the same group extended it to lower transmit frequencies (4.2–5.4 MHz) in ex vivo muscle with focused-ultrasound heating, motion compensation, and calibrated temperature maps, and to real-time closed-loop control in ex vivo tissue and anesthetized mice at 21 MHz imaging. What remains unresolved is whether the method survives clinically realistic conditions: heterogeneous, fat-containing tissue in which harmonic generation and its temperature dependence change across acoustic-property layers; deeper penetration at mid-range transmit frequencies (≈8 MHz) with longer focal lengths; and in vivo physiological (respiratory/cardiac) motion — together with the accuracy and precision of calibrated temperature maps in those settings. The problem is to establish and quantify the conditions under which harmonic CBE yields a calibrated, motion-robust thermometer in heterogeneous tissue at clinically relevant depth.

Background

Thermal therapies such as hyperthermia (heating tissue to roughly 40–45 °C, often to trigger drug release from thermosensitive carriers) and high-intensity focused ultrasound (HIFU) ablation require noninvasive temperature monitoring to control the delivered thermal dose. Magnetic resonance thermometry is the clinical standard but is expensive and requires MRI-compatible heating applicators; invasive probes give only single-point measurements. Ultrasound is a low-cost, real-time alternative. The classical ultrasound approach is the echo-shift technique: shifts in the arrival time of backscattered radio-frequency (RF) echo signals, caused by temperature-dependent sound speed and thermal expansion, are tracked by cross-correlation between frames. Its main weakness is extreme sensitivity to tissue motion (respiratory and cardiac), which masks the thermal shifts. A second approach is change in backscattered energy (CBE): the energy of the backscattered RF echo changes with temperature because the attenuation and backscatter coefficients of tissue are temperature-dependent (Arthur et al., 2003). CBE is comparatively less sensitive to motion, since it depends on echo amplitude rather than arrival time. A third ingredient is nonlinear harmonic generation: at high acoustic pressures, wave steepening in tissue generates harmonics (integer multiples) of the transmit frequency, whose amplitudes depend on the medium's coefficient of nonlinearity B/AB/A, attenuation, and sound speed — all of which are themselves temperature-dependent (Maraghechi et al., 2016, who simulated this temperature dependence in water at transmit frequencies from 1 to 20 MHz and measured it at 3.3 and 13 MHz). Maraghechi, Kolios and Tavakkoli (2019) combined these ideas into hCBE — CBE computed on the second- and third-harmonic bands rather than the fundamental — and showed in a tissue-mimicking gel phantom and ex vivo bovine muscle (13 MHz transmit, harmonics detectable up to ~39 MHz, local heating by hot water circulated through a needle, 26–46 °C) that harmonic CBE maps localize the heated region and remain usable under mechanically induced vibration where echo-shift maps fail. Higher harmonics showed higher temperature sensitivity. The authors posed a research agenda for clinical translation: the short focal length and high transmit frequency limit penetration depth, so lower transmit frequencies and longer focal lengths need investigation, with mid-range frequencies (≈ 8 MHz) suggested as potentially suitable since harmonic generation becomes difficult at low frequencies (≈ 1 MHz); tissue heterogeneities, especially fat, and varying source pressure amplitude will change harmonic generation at different temperatures; a focused therapeutic ultrasound transducer should replace needle heating as the heating modality; calibration maps converting measured harmonic CBE to temperature are needed, and the accuracy and precision of the resulting temperature maps must be established; and motion compensation combined with high-frame-rate acquisition should be demonstrated. Follow-up work by the same group resolved several agenda items: Shaswary et al. (2021, Ultrasonics) demonstrated second-harmonic CBE at a 4.2 MHz linear array (second harmonic ≈ 8.4 MHz) with low-intensity focused ultrasound heating of ex vivo pork muscle, a motion-compensation algorithm, and calibrated two-dimensional temperature maps over 37–47 °C; Hornsby, Shaswary and Tavakkoli (2021) improved the signal-to-noise ratio (SNR) of hCBE maps at 4.31 and 5.43 MHz via nonlinear frequency compounding; and Nguyen et al. (2024) achieved real-time closed-loop proportional-integral-derivative (PID) temperature control in ex vivo porcine tissue and in vivo anesthetized mouse hind leg (21 MHz imaging, 2 MHz focused ultrasound heating, 2D cross-correlation motion compensation), with CBE-estimated focal temperatures within about ±0.85 °C (ex vivo) and ±0.97 °C (in vivo) of a calibrated thermocouple. However, all of this was performed in homogeneous muscle (or phantoms) at shallow depth, and Nguyen et al. (2024) explicitly state that effective tissue motion compensation remains one of the main limitations to be addressed. No published study has examined harmonic CBE in heterogeneous, fat-containing media, at clinically relevant depth with longer focal lengths at mid-range transmit frequencies (≈ 8 MHz), or under true respiratory/cardiac motion.

Problem Statement

Establish whether ultrasound thermometry based on the change in backscattered energy of nonlinearly generated acoustic harmonics (hCBE) remains a valid, calibratable, and motion-robust temperature indicator under clinically realistic conditions, and quantify the accuracy and precision of the calibrated temperature maps in those conditions. Concretely, extending the demonstrated ex vivo homogeneous-muscle and anesthetized small-animal results to the regime the source work identified as open: (1) heterogeneous media containing fat and complex acoustic-property variation — determine how harmonic generation and its temperature dependence are altered across acoustic-property layers, whether a CBE-to-temperature calibration obtained under one condition transfers, and how source pressure amplitude affects the harmonic CBE–temperature relation; (2) mid-range transmit frequencies (≈ 8 MHz) with longer focal lengths for deeper clinical penetration — determine whether temperature-dependent harmonic CBE retains sufficient sensitivity and map quality (contrast-to-noise and signal-to-noise ratio) at depth, given that harmonic generation weakens as transmit frequency decreases; and (3) in vivo physiological motion (respiratory and cardiac, including elevational and radiation-force-induced components) — determine whether harmonic CBE combined with motion compensation and high-frame-rate acquisition produces reliable temperature maps in vivo. The overarching deliverable is a quantitative characterization: for each of these conditions, how the validity of harmonic CBE thermometry and the accuracy and precision of its calibrated temperature maps against independent ground truth depend on the governing parameters of that condition (for example, fat-layer geometry and acoustic contrast, penetration depth and transmit frequency, source pressure amplitude, and motion amplitude and frequency), including characterized failure modes delimiting where the method stops being valid. A single demonstration at one chosen parameter point — whether a successful temperature map or one characterized failure mode — does not by itself quantify the conditions under which harmonic CBE yields a calibrated, motion-robust thermometer. A resolving answer covers all three conditions and, for each, reports the validity of harmonic CBE thermometry and the accuracy and precision of its calibrated temperature maps as functions of at least the governing parameters listed above, together with at least one characterized failure boundary delimiting where the method stops being valid.

References

  1. Borna Maraghechi, Michael C. Kolios, Jahan Tavakkoli (2019). Feasibility of detecting change in backscattered energy of acoustic harmonics in locally heated tissues. International Journal of Hyperthermia 36(1):963–973. DOI: 10.1080/02656736.2019.1660001. URL: https://doi.org/10.1080/02656736.2019.1660001
  2. R. Martin Arthur, William L. Straube, Jared D. Starman, Eduardo G. Moros (2003). Noninvasive temperature estimation based on the energy of backscattered ultrasound. Medical Physics 30(6):1021–1029. DOI: 10.1118/1.1570373. URL: https://doi.org/10.1118/1.1570373
  3. Borna Maraghechi, Mojtaba H. Hasani, Michael C. Kolios, Jahan Tavakkoli (2016). Temperature dependence of acoustic harmonics generated by nonlinear ultrasound wave propagation in water at various frequencies. The Journal of the Acoustical Society of America 139(5):2475–2481. DOI: 10.1121/1.4946898. URL: https://doi.org/10.1121/1.4946898
  4. Elyas Shaswary, Hisham Assi, Celina Yang, J. Carl Kumaradas, Michael C. Kolios, Gholam Peyman, Jahan Tavakkoli (2021). Noninvasive calibrated tissue temperature estimation using backscattered energy of acoustic harmonics. Ultrasonics 114:106406. DOI: 10.1016/j.ultras.2021.106406. URL: https://doi.org/10.1016/j.ultras.2021.106406
  5. Tyler Hornsby, Elyas Shaswary, Jahan Tavakkoli (2021). Development of an ultrasonic nonlinear frequency compounding method with applications in tissue thermometry. The Journal of the Acoustical Society of America 150(4):3192–3203. DOI: 10.1121/10.0006207. URL: https://doi.org/10.1121/10.0006207
  6. Michael Nguyen, Na Zhao, Yuan Xu, Jahan Tavakkoli (2023). Decorrelated compounding of synthetic aperture ultrasound imaging to detect low contrast thermal lesions induced by focused ultrasound. Ultrasonics 134:107098. DOI: 10.1016/j.ultras.2023.107098. URL: https://doi.org/10.1016/j.ultras.2023.107098
  7. Michael Nguyen, Ayushi Agarwal, J. Carl Kumaradas, Michael C. Kolios, Gholam Peyman, Jahan Tavakkoli (2024). Real-time non-invasive control of ultrasound hyperthermia using high-frequency ultrasonic backscattered energy in ex vivo tissue and in vivo animal studies. Physics in Medicine and Biology 69:215001. DOI: 10.1088/1361-6560/ad7f19. URL: https://doi.org/10.1088/1361-6560/ad7f19
  8. Adrien Rohfritsch, Victor Barrere, Laura Estienne, David Melodelima (2024). 2D ultrasound thermometry during thermal ablation with high-intensity focused ultrasound. Ultrasonics 142:107372. DOI: 10.1016/j.ultras.2024.107372. URL: https://doi.org/10.1016/j.ultras.2024.107372
  9. Kun Yang, Qiang Li, Xiaowei Zhou, Chiao-Yin Wang, Po-Hsiang Tsui (2024). Ultrasound Delta CBE Imaging: A New Approach Based on Local Energy Subtraction to Localization of the HIFU Focal Spot Using Changes in Backscattered Energy. Journal of Medical and Biological Engineering 44(4):618–627. DOI: 10.1007/s40846-024-00887-3. URL: https://doi.org/10.1007/s40846-024-00887-3