Browsing by Author "Haqshenas, Reza"

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    Classifying acoustic cavitation with machine learning trained on multiple physical models
    (2024) Gatica González, Trinidad Anastasia; Wout, Elwin van't; Haqshenas, Reza
    Acoustic cavitation refers to the formation and oscillation of microbubbles in a liquid exposed to acoustic waves. Depending on the properties of the liquid and the parameters of the acoustic waves, bubbles behave differently. The two main regimes of bubble dynamics are transient cavitation, where a bubble collapses violently, and stable cavitation, where a bubble undergoes periodic oscillations. Predicting these regimes under specific sonication conditions is important in biomedical ultrasound and sonochemistry. For these predictions to be helpful in practical settings, they must be precise and computationally efficient. In this study, we have used machine learning techniques to predict the cavitation regimes of air bubble nuclei in a liquid. The supervised machine learning was trained by solving three differential equations for bubble dynamics, namely the Rayleigh-Plesset, Keller-Miksis, and Gilmore equations. These equations were solved for a range of initial parameters, including temperature, bubble radius, acoustic pressure, and frequency. Four different classifiers were developed to label each simulation as either stable or transient cavitation. Subsequently, four different machine-learning strategies were designed to analyze the likelihood of transient or stable cavitation for a given set of acoustic and material parameters. Cross-validation on held-out test data shows a high accuracy of the machine learning predictions. The results indicate that machine learning models trained on physics-based simulations can reliably predict cavitation behavior across a wide range of conditions relevant to real-world applications. This approach can be employed to optimize device settings and protocols used in imaging, therapeutic ultrasound, and sonochemistry.
  • Thumbnail Image
    Item
    Evaluation of fetal exposure to environmental noise using a computer-generated model
    (Research Square, 2024) Gélat, Pierre; Wout, Elwin van't; Haqshenas, Reza; Melbourne, Andrew; David, Anna; Mufti, Anna; Henriques, Julian; Thibaut de Maisieres, Aude; Jauniaux, Eric
    Acoustic noise can have profound effects on wellbeing, impacting the health of the pregnant mother and the development of the fetus. Mounting evidence suggests neural memory traces are formed by auditory learning in utero. A better understanding of the fetal auditory environment is therefore critical to avoid exposure to damaging noise levels. Using anatomical data from MRI scans (𝑁 = 3), we used a computational model to quantify the acoustic field inside the pregnant maternal abdomen. We obtained acoustic transfer characteristics across the human audio range and pressure maps in transverse planes passing through the uterus at 5 kHz, 10 kHz and 20 kHz, showcasing multiple scattering and modal patterns. Our calculations suggest that for all datasets, the sound transmitted in utero is attenuated by as little as 6 dB below 1 kHz, confirming results from animal studies that the maternal abdomen and pelvis do not shelter the fetus from external noise.
  • No Thumbnail Available
    Item
    Full-Wave Modeling of Transcranial Ultrasound using Volume-Surface Integral Equations and CT-Derived Heterogeneous Skull Data
    (2025) Almuna Morales, Alberto; Aballay Fernandez, Danilo Ignacio; Gélat, Pierre; Haqshenas, Reza; Vant Wout, Elwin
    Transcranial ultrasound therapy uses focused acoustic energy to induce therapeutic bioeffects in the brain. Ultrasound is transmitted through the skull, which is highly attenuating and heterogeneous, causing beam distortion, reducing focal pressure, and shifting the target location. Computational models are frequently used for predicting beam aberration, assessing cranial heating, and correcting the phase of ultrasound transducers. These models often rely on computed tomography (CT) images to build patient-specific geometries and estimate skull acoustic properties. However, the coarse voxel resolution of CT limits accuracy for differential equation solvers. This paper presents an efficient numerical method based on volume-surface integral equations to model full-wave acoustic propagation through heterogeneous skull tissue. We have shown that this approach is highly accurate on relatively coarse meshes compared to the minimum wavelength, enabling direct use of CT voxel data. The method is validated against a high-resolution boundary element model using an averaged skull representation. Simulations with a CT-based skull model and a bowl transducer show significant beam distortion and attenuation, with a focal shift of several millimeters from the homogeneous case