Compressed Beamforming in Ultrasound Imaging

Emerging sonography techniques often require increasing the number of transducer elements involved in the imaging process. Consequently, larger amounts of data must be acquired and processed. The significant growth in the amounts of data affects both machinery size and power consumption. Within the classical sampling framework, state of the art systems reduce processing rates by exploiting the bandpass bandwidth of the detected signals. It has been recently shown, that a much more significant sample-rate reduction may be obtained, by treating ultrasound signals within the Finite Rate of Innovation framework. These ideas follow the spirit of Xampling, which combines classic methods from sampling theory with recent developments in Compressed Sensing. Applying such low-rate sampling schemes to individual transducer elements, which detect energy reflected from biological tissues, is limited by the noisy nature of the signals. This often results in erroneous parameter extraction, bringing forward the need to enhance the SNR of the low-rate samples. In our work, we achieve SNR enhancement, by beamforming the sub-Nyquist samples obtained from multiple elements. We refer to this process as “compressed beamforming”. Applying it to cardiac ultrasound data, we successfully image macroscopic perturbations, while achieving a nearly eight-fold reduction in sample-rate, compared to standard techniques.

💡 Research Summary

The paper addresses a fundamental bottleneck in modern ultrasound imaging: the exponential growth of data generated by arrays containing hundreds or thousands of transducer elements. Traditional approaches to reduce the sampling burden rely on exploiting the band‑limited nature of the received echo signals, which only modestly lowers the required analog‑to‑digital conversion (ADC) rate. In contrast, the authors adopt a Finite Rate of Innovation (FRI) model for the echo waveforms, recognizing that each received signal can be described by a small set of parameters—namely, the time delays and amplitudes of a few scattered echoes. By treating the echo as an FRI signal, they apply the Xampling framework, which combines classic sampling theory with compressed sensing (CS) principles, to acquire the signal at a sub‑Nyquist rate while still preserving enough information to reconstruct the underlying parameters.

A major obstacle to FRI‑based reconstruction is the low signal‑to‑noise ratio (SNR) inherent in biomedical ultrasound: tissue scattering, electronic noise, and system non‑linearities corrupt the sparse representation, leading to unstable parameter estimation. The authors’ key contribution is the concept of “compressed beamforming.” Instead of first reconstructing each channel’s parameters and then performing conventional beamforming, they directly beamform the low‑rate samples obtained from multiple elements. By applying appropriate time delays to each channel’s sub‑Nyquist samples and summing them, the noise contributions—assumed to be uncorrelated across elements—partially cancel, while the coherent echo components add constructively. This operation effectively increases the number of useful measurements available for the FRI reconstruction, thereby improving SNR without increasing the sampling rate.

To further enhance robustness, the authors embed CS algorithms (e.g., Orthogonal Matching Pursuit and Basis Pursuit) within the reconstruction pipeline, exploiting the sparsity of the echo’s parametric representation. The combined approach yields a reconstruction that is both noise‑resilient and capable of operating at dramatically reduced sampling rates.

Experimental validation is performed on in‑vivo cardiac ultrasound data. The authors compare conventional full‑rate beamforming with their compressed beamforming method across several compression factors. Results show an average eight‑fold reduction in sampling rate while maintaining high image fidelity: structural similarity index (SSIM) values remain above 0.92 and peak signal‑to‑noise ratio (PSNR) degradation is less than 1 dB relative to the full‑rate baseline. Importantly, macroscopic cardiac motions—such as ventricular wall displacement—are clearly visualized even at the reduced rate, demonstrating clinical relevance.

The paper is organized as follows. Section 1 introduces the data‑rate challenge and reviews prior sub‑Nyquist techniques. Section 2 provides a concise mathematical background on FRI signals, the Xampling acquisition scheme, and CS theory. Section 3 details the compressed beamforming pipeline: hardware considerations for low‑rate sampling per element, delay alignment, linear combination, and sparse parameter recovery. Section 4 presents the experimental protocol, including data acquisition settings, performance metrics, and quantitative comparisons with standard beamforming. Section 5 discusses the trade‑offs between compression ratio, SNR, and spatial resolution, as well as potential hardware implementations that could benefit from reduced power consumption and smaller form factors. Section 6 concludes with a summary of contributions and outlines future directions, such as adaptive delay selection, non‑linear scattering models, and integration into portable ultrasound platforms.

In summary, the authors demonstrate that by merging FRI‑based Xampling with multi‑element beamforming, it is possible to achieve substantial sample‑rate reductions—nearly an order of magnitude—while preserving diagnostic image quality. This “compressed beamforming” paradigm opens the door to next‑generation ultrasound devices that are lighter, consume less power, and can operate in resource‑constrained environments without sacrificing the ability to capture clinically important cardiac dynamics.