Hassan Rivaz

The frequency dependence of backscattered radiofrequency (RF) signals produced by ultrasound scanners carries rich information related to the tissue microstructure (i.e., scatterer size, attenuation). This information can be sue to classify tissues based on microstructural changes associated to disease onset and progression. Conventional convolutional neural networks (CNNs) can learn this information directly from radio-frequency (RF) data, but they often struggle to achieve adequate frequency selectivity. This increases model complexity and convergence time, and limits generalization. To overcome these challenges, SincNet, originally developed for speech processing, was adapted to classify RF data based on differences in frequency properties. Rather than learning every filter coefficient, SincNet only learns each filter's low frequency and bandwidth, dramatically reducing the number of parameters and improving frequency resolution. For model interpretability, a Gradient-Weighted Filter Contribution is introduced, which highlights the importance of spectral bands. The approach was validated on three datasets: simulated data with different scatterer sizes, experimental phantom data, and in vivo data of rats which were fed a methionine and choline- deficient diet to develop liver steatosis, inflammation, and fibrosis. The modified SincNet consistently achieved the best results in material/tissue classifications.

Accurate and generalizable object segmentation in ultrasound imaging remains a significant challenge due to anatomical variability, diverse imaging protocols, and limited annotated data. In this study, we propose a prompt-driven vision–language model (VLM) that integrates grounding DINO with SAM2 to enable object segmentation across multiple ultrasound organs. A total of 18 public ultrasound datasets, encompassing the breast, thyroid, liver, prostate, kidney, and paraspinal muscle, were utilized. These datasets were divided into 15 for fine-tuning and validation of grounding DINO using low-rank adaptation (LoRA) to the ultrasound domain, and three were held out entirely for testing to evaluate performance in unseen distributions. Comprehensive experiments demonstrate that our approach outperforms state-of-the-art (SOTA) segmentation methods, including UniverSeg, MedSAM, MedCLIP-segment anything model (SAM), BiomedParse, and SAMUS on most seen datasets while maintaining strong performance on unseen datasets without additional fine-tuning. These results underscore the promise of VLMs in scalable and robust ultrasound image analysis, reducing dependence on large, organ-specific annotated datasets. We will publish our code on code.sonography.ai after acceptance.

Ultrasound Coherent Plane Wave Compounding (CPWC) enhances image contrast by combining echoes from multiple steered transmissions. While increasing the number of angles generally improves image quality, it drastically reduces the frame rate and can introduce blurring artifacts in fast-moving targets. Moreover, compounded images remain susceptible to noise, particularly when acquired with a limited number of transmissions. We propose a zero-shot denoising framework tailored for low-angle CPWC acquisitions, which enhances contrast without relying on a separate training dataset. The method divides the available transmission angles into two disjoint subsets, each used to form compound images that include higher noise levels. The new compounded images are then used to train a deep model via a self-supervised residual learning scheme, enabling it to suppress incoherent noise while preserving anatomical structures. Because angle-dependent artifacts vary between the subsets while the underlying tissue response is similar, this physics-informed pairing allows the network to learn to disentangle the inconsistent artifacts from the consistent tissue signal. Unlike supervised methods, our model requires no domain-specific fine-tuning or paired data, making it adaptable across anatomical regions and acquisition setups. The entire pipeline supports efficient training with low computational cost due to the use of a lightweight architecture, which comprises only two convolutional layers. Evaluations on simulation, phantom, and in vivo data demonstrate superior contrast enhancement and structure preservation compared to both classical and deep learning-based denoising methods.

Shear wave elastography (SWE) provides quantitative markers for tissue characterization by measuring shear wave speed (SWS), which reflects tissue stiffness. SWE uses an acoustic radiation force pulse sequence to generate shear waves that propagate laterally through tissue with transient displacements. These waves travel perpendicular to the applied force, and their displacements are tracked using high-frame-rate ultrasound. Estimating the SWS map involves two main steps: speckle tracking and SWS estimation. Speckle tracking calculates particle velocity by measuring RF/IQ data displacement between adjacent firings, while SWS estimation methods typically compare particle velocity profiles of samples that are laterally a few millimeters apart. Deep learning (DL) methods have gained attention for SWS estimation, often relying on supervised training using simulated data. However, these methods may struggle with real-world data, which can differ significantly from simulated training data, potentially leading to artifacts in the estimated SWS map. To address this challenge, we propose a physics-inspired learning approach that utilizes real data without known SWS values. Our method employs an adaptive unsupervised loss function, allowing the network to train with real noisy data to minimize the artifacts and improve the robustness. We validate our approach using experimental phantom data and in vivo liver data from two human subjects, demonstrating enhanced accuracy and reliability in SWS estimation compared to conventional and supervised methods. This hybrid approach leverages the strengths of both data-driven and physics-inspired learning, offering a promising solution for more accurate and robust SWS mapping in clinical applications.

Ultrasound imaging is crucial in medical diagnostics, offering real-time visualization of internal anatomical structures. However, accurate automatic segmentation of ultrasound images remains challenging, particularly in scenarios with limited labeled data. In this paper, we propose a semi-supervised learning approach for ultrasound image segmentation, leveraging the statistics of data in unlabeled images to enhance segmentation accuracy. Our method builds upon the encoder-decoder architecture and incorporates innovative semi-supervised learning techniques based on contrastive learning. We have collected ultrasound images from 80 patients and 34 healthy volunteers, focusing on applications in sarcopenia assessment and emergency response scenarios. We demonstrate the effectiveness of our approach through extensive experiments on expert segmentations in this dataset.Our results demonstrate the superior performance of the proposed method across various training data splits (i.e., 1%, 5%, 10%, 20%, 30%, and 100%). While U-NET performed the best with 100% of the training data (i.e., 154 annotated images), the proposed method achieved comparable performance with only 10% of the data (i.e., 16 annotated images). Furthermore, statistical analysis confirmed that our method significantly outperforms existing models, including U-NET, CCT, and UniMatch, in most scenarios (i.e., training set splits). These findings highlight the robustness and efficiency of the proposed method, especially in environments where labeled data is scarce

Most modern models designed for ultrasound (US) image segmentation are characterized by high computational and memory requirements, limiting their practical utility in point-of-care US settings. Consequently, researchers have devised innovative approaches to compress these large models, enabling the training of smaller networks capable of achieving comparable generalization performance. Among these strategies, knowledge distillation (KD) has emerged as particularly suitable for scenarios involving small datasets or where significant efficiency improvements are desired. While previous KD-based methods have focused on extracting comprehensive information from diverse levels of teacher representation, they often overlook the identification of the most effective representation level. Additionally, many existing techniques propose intricate strategies that present implementation challenges. To address this gap, our study concentrates on selecting optimal teacher representations from various levels. Through an exhaustive analysis of KD pathways, loss functions, and the impact of augmentation, we offer valuable insights into the mechanisms underlying knowledge transfer from the teacher to the student networks. Our proposed methodology significantly enhances student performance, elevating the Dice similarity score from 73% to 80%, while the teacher model achieves 81%. Notably, our student model achieves this improvement with only 0.82 million parameters, compared to the teacher model’s 96 million parameters.

The ultrasound backscatter coefficient is a frequency-dependent quantity intrinsic to biological tissues that can be recovered from backscattered radiofrequency signals, granted acquisitions on a reference phantom are available under the same system’s settings. A phantom-free backscatter coefficient estimation method is proposed based on Gaussian-shaped approximation of the point spread function (electronics and piezoelectric characteristics of the scanner’s probe) and the effective medium theory combined with the structure factor model, albeit the proposed approach is amenable to other models. Meanwhile, the total attenuation due to intervening tissues is refined from its theoretical value, which is based on reported average behaviors of tissues, while allowing correction for diffraction due to the probe’s geometry. The reference phantom method adapted to a similar approach except for the Gaussian approximation is also presented. The proposed phantom-free and reference phantom methods were compared on ten COVID-19 positive patients and twelve control subjects with measures on femoral veins and arteries. In this context, red blood cells are viewed as scatterers that form aggregates increasing the backscatter under the COVID-19 inflammatory condition. The considered model comprises five parameters, including the mean aggregate size estimated according to polydispersity of aggregates’ radii, and anisotropy of their shape. The mean aggregate size over the two proposed methods presented an intraclass correlation coefficient of 0.964 for consistency. The aggregate size presented a significant difference between the two groups with either two methods, despite the confounding effect of the maximum Doppler velocity within the blood vessel and its diameter.