Novel Application of Acoustic Beamforming for Non-Invasive Profiling of Cardiac Fibrosis

All figures and diagrams were created by the student researcher unless otherwise noted.

Problem: Non-invasive Profiling of Cardiac Fibrosis

Cardiac fibrosis contributes to heart failure in millions of patients and is driven by excessive fibroblast activation, leading to stiffened myocardium and disrupted cardiac function. Emerging therapies such as fibroblast-targeted CAR-T cells show promise in reversing fibrosis, but there is currently no non-invasive method to directly quantify and localize fibroblast clearance.

Current techniques like tissue biopsies through histological staining are invasive and low-resolution, while imaging such as echocardiography fail to directly quantify fibrosis.

Solution: Acoustic Beamforming

This project introduces a novel, non-invasive MEMS microphone array system that applies acoustic beamforming and chaos-based signal analysis to detect and localize cardiac fibrosis.

The approach is based on three key ideas:

Fibrotic tissue produces irregular, high-frequency turbulent acoustic signatures due to stiffness and scattering when turbulent blood flow interacts with heterogeneous tissue, as seen with S3/S4 vibrations in cardiac cycle.

Healthy tissue produces periodic, low-frequency signals during systole and diastole with minimal chaos in sound amplitudes.

These differences can be spatially resolved using beamforming to localize cardiac fibrosis to given voxel locations and quantified through a novel turbulence power index.

By converting chaotic cardiac vibrations through a chaos-theory-inspired methodology into heatmaps of tissue health, this system enables non-invasive profiling of fibrosis and its regression following anti-fibrotic therapeutics.

Research Question:

Can acoustic beamforming combined with chaotic signal analysis be used to non-invasively localize and quantify cardiac fibrosis?

Importance:

Eliminates need for biopsy, radiation, or contrast agents

Provides a pathway toward personalized treatment through drug discovery in patients

Methodology:

Phase 1: Myocardial Tissue Modeling (Hydrogels)

Alginate-gelatin hydrogels were engineered to model myocardial tissue: (1) healthy, (2) fibrotic, and (3) post-CAR-T tissue. Controlled CaCl₂ crosslinking altered stiffness and porosity to match physiological ranges. Internal gelation ensured uniform structure. Microstructural validation using SEM confirmed physiological ranges through porosity. Non-conductive samples were sputter coated at 10nm of gold/palladium, mounted using carbon tape, and analysed through an SEM at the University of Ottawa.

Phase 2: Acoustic Phantom Construction & PCB for Data Analysis

An acoustic phantom mimicking the human torso was constructed for all three hydrogel conditions. Simulated cardiac sounds (20–500 Hz range) were played through the system to replicate hemodynamic vibrations during cardiac cycle, specifically turbulence in S3/S4 waveforms. 5 voxel positions were assigned and a each voxel position. The same standardized cardiac waveform was played as all 3 MEMS microphones simultaneously recorded the transmitted and scattered acoustic response from a localized source point.

A MEMS microphone array (≈ 5 mm spacing) connected to an ESP-32 captured these distinct acoustic signals. A custom printed circuit board (PCB) design was developed to support multi-point data acquisition for future clinical translation. The PCB was designed using schematics with 3 MEMS microphones connected to an analog-to-digital converter (ADC IC) which processes audio waveforms and converts them into digital values that the ESP32 can process. Copper tracing was then completed in the KICAD software to render manufacturing files.

Phase 3: Acoustic Signal Processing (Beamforming)

Recorded signals were transformed using Short-Time Fourier Transform (STFT), applied voxel weights, and processed via delay-and-sum beamforming to isolate voxel-specific vibrations. Key features were extracted: (1) S3 amplitude, (2) S4 amplitude, and (3) Lyapunov exponent (λ) to quantify signal chaos. These variables were normalized and combined into a turbulence power index (TPI), representing the degree of acoustic ‘chaos’ associated with fibrosis.

Results:

1. Acoustic Model Differentiates Fibrotic States

Simulated cardiac sound files spanning 20–500 Hz:  synthesized S1/S2 carrier (60–100 Hz), S3 gallop (25–70 Hz), and S4 presystolic (20–50 Hz) components were played through the phantom across 5 voxel positions. Distinct acoustic signatures were observed across all three tissue states: fibrotic phantoms produced high-frequency, chaotic waveforms consistent with chaotic scattering from dense collagen networks, while healthy phantoms showed stable, low-frequency signals. Post-CAR-T samples demonstrated heterogenous waveform patterns, indicating partial structural restoration of the extracellular matrix.

2. Turbulence Power Index (TPI) Quantifies Fibrosis

TPI values clearly separated conditions: healthy (0.11–0.15), fibrotic (0.85–0.91), and post-CAR-T (0.38–0.44). This demonstrates that combining S3 amplitude, S4 amplitude, and Lyapunov exponent provides a reliable quantitative biomarker for fibrosis severity.

3. Beamforming Enables Localization

Delay-and-sum beamforming successfully reconstructed voxel-specific waveforms, allowing identification of localized high-turbulence regions. Heatmaps revealed fibrotic zones, demonstrating that multi-point auscultation can resolve spatial heterogeneity in tissue properties.

4. SEM Validates Microstructural Differences

SEM analysis confirmed reduced porosity in fibrotic hydrogels (~12.5%) compared to healthy (~24.7%), with post-CAR-T samples showing intermediate values (~18.5%). These structural differences directly align with postulated acoustic scattering patterns.

Challenges and Limitations

While the acoustic system successfully localizes fibrotic regions, spatial resolution is limited by the small number of microphones. Additionally, validation was conducted using an in vitro hydrogel phantom rather than biological tissue, meaning physiological complexity is not fully captured. Future work will involve larger MEMS microphone sensor arrays and testing in organoid models to improve clinical translation.

Conclusions

This project establishes that acoustic beamforming combined with chaos signal analysis can non-invasively detect, localize, and quantify cardiac fibrosis in hydrogel tissue phantoms. The turbulence power index serves as a novel composite biomarker for tracking fibroblast clearance, with condition-level separation confirmed at p = 0.002.

Cost Analysis

The prototype system is highly cost-effective, with the PCB costing approximately $35 CAD. MEMS microphones and the  ESP-32 microcontrollers are inexpensive, making this approach significantly more accessible than imaging technologies like MRI or echocardiography. This system supports novel drug discovery of anti-fibrotic therapeutics.

Applications to Medicine and Society

This system enables personalized cardiac care by allowing clinicians to monitor fibrosis progression and treatment responses. It reduces reliance on invasive biopsies and expensive imaging, improving accessibility for patients. In drug development, it provides a platform to evaluate anti-fibrotic therapies more efficiently, widening innovation in cardiovascular therapies against cardiac fibrosis.

Future Work: Organoid Validation

A follow-up experiment will use cardiac organoids containing fibroblasts, cardiomyocytes, and extracellular matrix to replicate the positive feedback loop seen with both cell types. Fibrosis will be induced using TGF-β, followed by CAR-T treatment targeting fibroblast activation protein. The acoustic system will then track changes in turbulence power index over time. This model introduces cellular complexity providing a clinically relevant platform to validate whether acoustic signatures accurately reflect fibrosis regression.

The SEM work conducted for this experiment was funded and supervised by the Department of Materials Characterization at the University of Ottawa.  I would like to thank Dr. Yun Liu for helping facilitate the SEM imaging for this project and aiding in the sputter coating of my samples. I am also grateful to Dr. Eva Hemmer for her guidance on dry sample SEM imaging and for referring me to the Department of Materials Characterization at the University of Ottawa. Finally, I would like to thank Dr. Jianqun Wang from Carleton University for his advice on properly drying samples for SEM preparation.

Main Journal Articles:

Aghajanian H, Kimura T, Rurik JG, et al. Targeting cardiac fibrosis with engineered T cells. Nature. 2019;573(7774):430–433. doi:https://doi.org/10.1038/s41586-019-1546-z

Badawe HM, Raad P, Khraiche ML. High-resolution acoustic mapping of tunable gelatin-based phantoms for ultrasound tissue characterization. Frontiers in Bioengineering and Biotechnology. 2024;12. doi:https://doi.org/10.3389/fbioe.2024.1276143

Hinderer S, Schenke-Layland K. Cardiac fibrosis – A short review of causes and therapeutic strategies. Advanced Drug Delivery Reviews. 2019;146:77–82. doi:https://doi.org/10.1016/j.addr.2019.05.011

Kaberova Z, Karpushkin E, Nevoralová M, et al. Microscopic Structure of Swollen Hydrogels by Scanning Electron and Light Microscopies: Artifacts and Reality. Polymers. 2020;12(3):578. doi:https://doi.org/10.3390/polym12030578

Rathod RH, Powell AJ, Geva T. Myocardial Fibrosis in Congenital Heart Disease. Circulation Journal. 2016;80(6):1300–1307. doi:https://doi.org/10.1253/circj.cj-16-0353

Rurik JG, Tombácz I, Yadegari A, et al. CAR T cells produced in vivo to treat cardiac injury. Science. 2022;375(6576):91–96. doi:https://doi.org/10.1126/science.abm0594

Sato Y, Kawasaki T, Honda S, et al. Third and Fourth Heart Sounds and Myocardial Fibrosis in Hypertrophic Cardiomyopathy. Circulation Journal. 2018;82(2):509–516. doi:https://doi.org/10.1253/circj.cj-17-0650

Yeo TM, Chin C, Alvin W, et al. Global Prevalence of Myocardial Fibrosis among Individuals with Cardiometabolic Conditions: A Systematic Review and Meta-Analysis. European Journal of Preventive Cardiology. 2025;32(12). doi:https://doi.org/10.1093/eurjpc/zwaf083