This study compared two imaging methods—ultrasound and chest X-ray (CXR)—to see how well they assess whether fluid has been fully removed from the pleural space after thoracentesis, a procedure that drains fluid around the lungs. Patients with free-flowing pleural effusions were enrolled, and after the procedure, an ultrasound was performed immediately, while a CXR was done within four hours. The main outcome was whether both methods agreed on complete pleural space evacuation, defined as no remaining fluid seen on ultrasound or no blunting of the costophrenic angle on CXR.

A total of 145 patients were analyzed, with a median age of 64 years, and cancer was the most common cause of fluid buildup. The lung was considered trapped in half of the patients. More than 800 ultrasound images were reviewed independently by pulmonologists and radiologists who did not know the patient or procedure details. The results showed very strong agreement between ultrasound and CXR (agreement coefficient 0.93, 95% CI 0.83–1.00). When comparing based on effusion size, there was substantial agreement (kappa 0.64, 95% CI 0.51–0.77). Agreement between proceduralists and blinded ultrasound reviewers was also strong (kappa 0.81, 95% CI 0.71–0.90).

In conclusion, post-thoracentesis ultrasound is an equally effective alternative to chest X-ray for evaluating pleural space evacuation in patients with simple pleural effusions.

Figure 1.Pre and postthoracentesis evaluation of pleural effusion using ultrasound and chest radiography (CXR). (A) Large free-flowing simple pleural effusion visible in the posterior view. (B,��C) Effusion size measured in the anterior and midaxillary views. (D) Prethoracentesis CXR with a pleural effusion (arrow). (��–G) Post-thoracentesis images showing resolution of the effusion and complete pleural evacuation in the midaxillary, anterior, and posterior views. (H) Post-thoracentesis CXR with complete pleural evacuation.

Functional ultrasound was first tested in rodents, but more recently, it has been used in humans in situations where the skull can be bypassed, such as during surgery when the skull is removed or in newborns whose skulls have not yet fused. While these demonstrations are important, they do not allow for widespread clinical or scientific use in humans.������

In this study, we successfully performed transcranial functional ultrasound in 13 human subjects. To test its effectiveness, we used a breath-hold task. While we relied mostly on standard methods, we introduced two key improvements: direct axial motion correction and compound Barker coded excitation. Across all 13 participants, the functional ultrasound signal showed a correlation of 0.53±0.08 with the expected brain activity pattern, which is comparable to similar functional MRI (fMRI) studies.����

��

Deep neural network (DNN) beamformers are becoming increasingly popular in ultrasound imaging due to their ability to handle complex, non-linear functions. In the context of clutter suppression, these DNN beamformers are typically trained using synthetic data, where the model learns by comparing its predictions to a known ground truth.��

In this study, we found that the commonly used Smooth-L1 loss function does not necessarily lead to improvements in the generalized contrast-to-noise ratio (gCNR), a key measure of image quality. To address this, we introduced a new training approach that directly incorporates gCNR as a regularizer in the loss function. Our results showed that this method improved performance on synthetic data. Furthermore, when we integrated gCNR regularization into an existing domain adaptation technique, we achieved a measurable improvement in gCNR, with a gain of 0.0322 ± 0.0336 compared to the traditional delay-and-sum method.��

Ultrasound images are often processed using a method called delay-and-sum beamforming, which results in data following a predictable pattern (a Rayleigh distribution). Many traditional image quality measurements, like signal-to-noise ratio (SNR), were designed based on this pattern. However, newer ultrasound imaging techniques, such as advanced beamforming and post-processing methods, change the data in ways that no longer fit this traditional pattern. This makes older quality measurements less reliable. To solve this, researchers have developed improved metrics like the generalized contrast-to-noise ratio (gCNR), which can accurately assess image quality regardless of how the data has been transformed. In this study, we propose a new version of SNR that incorporates gCNR and other modern techniques. This updated measurement will be more reliable across different ultrasound imaging methods while still maintaining the important clinical insights that SNR was originally designed to provide.��

High-frequency ultrasound (HFUS) is a safe and effective method for visualizing the characteristics of skin tumors, including how deep they are, which is important for planning treatment. Researchers wanted to see how well HFUS measures the depth of melanoma (a type of skin cancer) compared to the current best method, which involves examining the tumor tissue under a microscope (histopathology).

They reviewed 36 studies that used HFUS with frequencies of at least 10 MHz to measure tumor depth and then compared those measurements to histopathology results. The studies included various types of melanoma, mostly found on the trunk and limbs. The ultrasound frequencies used ranged from 13 MHz to 100 MHz, and the number of participants in the studies ranged from 5 to 264.

The correlation between HFUS and histopathology in measuring tumor depth varied, with higher frequency probes (≥70 MHz) showing the best accuracy. Specifically, lower frequency probes (10-20 MHz) were less accurate. Melanomas that were thicker (more than 0.75 mm deep) were measured more accurately by HFUS. Some reasons HFUS might show a deeper measurement than histopathology include immune cells in the tumor, presence of a mole, and tissue shrinkage during processing.

The researchers also noted that many studies did not report important details about the tumors they measured. Despite these gaps, HFUS could be a useful additional tool for assessing melanoma before surgery, especially for thicker tumors, with higher frequencies providing better accuracy.

Previous research demonstrated that domain adaptive deep neural networks (DNNs) can surpass delay-and-sum (DAS) beamforming in abdominal imaging. This study explores the potential of applying the domain adaptive DNN framework to transthoracic echocardiography (TTE). Additionally, architectural enhancements such as using an encoder-decoder structure and skip connections are proposed to improve ultrasound image quality for tasks like detecting thrombi in the left atrial appendage (LAA).

The DNN training data comprised both simulated and in vivo cardiac data. Simulated anechoic and hypoechoic cysts with varying levels of clutter were created using Field II, while in vivo data were obtained from patients at ������ý���� Medical Center. Fundamental frequency TTE data from five separate cases were processed using DAS, ADMIRE, the baseline model, and several models with modified architectures.

The study found that, regardless of the amount of training data, DNNs consistently achieved higher generalized contrast-to-noise ratio (gCNR) and contrast ratio (CR) but lower contrast-to-noise ratio (CNR) compared to DAS. The best-performing beamformer was a DNN with the proposed architectural improvements, achieving average gCNR and CR values of 0.907 and 48.30 dB, respectively, compared to the baseline DNN values of 0.788 and 39.45 dB, and DAS values of 0.717 and 14.08 dB.

These results indicate that the domain adaptive DNN framework is effective for transthoracic cardiology applications. Moreover, the encoder-decoder architecture with skip connections can further enhance image quality. Future advancements may yield even greater improvements in ultrasound imaging quality.

Early image quality metrics were often developed with the goal of aligning with clinicians’ subjective opinions on what constitutes better images. As imaging techniques advanced, especially with the introduction of adaptive beamformers and other post-processing methods, these older metrics often failed to remain accurate. This disconnect allowed some beamformers to manipulate these metrics without actually improving clinical image quality.

In this work, a metric known as the Signal-to-Noise Ratio (SNR) for lesion detectability, originally proposed by Smith et al., is examined and improved. The new version, called generalized SNR (gSNR), incorporates the generalized Contrast-to-Noise Ratio (gCNR) to provide more reliable assessments that resist manipulation. Unlike gCNR, the original SNR includes lesion size and spatial resolution in its calculations, which enhances its robustness.

The paper analytically demonstrates that for Rayleigh-distributed data, gCNR can be expressed in terms of another metric, Cψ, proposed by Smith et al. This allows gCNR to substitute for SNR while also considering more accurate methods for estimating resolution cell size. Consequently, the gSNR provides a more reliable measure of lesion detectability that better aligns with clinical evaluations of image quality.

Ultrasound imaging is widely used in medicine due to its safety and cost-effectiveness compared to other methods. However, its quality can vary depending on tissue properties and depth. In this study, researchers tested deep learning techniques to create 3D models of objects imaged with ultrasound. They used three versions of the 3D U-Net model, each trained with different scenarios. The models performed well on specific categories of objects they were trained on but struggled with new categories. Researchers also looked into dual-task autoencoding to improve performance across different object types. These findings set a foundation for further improving the U-Net model to handle a broader range of ultrasound imaging tasks, potentially enhancing visualization and accuracy in medical applications.