Background: Ultrasound imaging is the most frequently performed for the patients with chronic hepatitis or liver cirrhosis. However, ultrasound imaging is highly operator dependent and interpretation of ultrasound images is subjective, thus well-trained radiologist is required for evaluation. Automated classification of liver fibrosis could alleviate the shortage of skilled radiologist especially in low-to-middle income countries. The purposed of this study is to evaluate deep convolutional neural networks (DCNNs) for classifying the degree of liver fibrosis according to the METAVIR score using US images. Methods: We used ultrasound (US) images from two tertiary university hospitals. A total of 7920 US images from 933 patients were used for training/validation of DCNNs. All patient were underwent liver biopsy or hepatectomy, and liver fibrosis was categorized based on pathology results using the METAVIR score. Five well-established DCNNs (VGGNet, ResNet, DenseNet, EfficientNet and ViT) was implemented to predict the METAVIR score. The performance of DCNNs for five-level (F0/F1/F2/F3/F4) classification was evaluated through area under the receiver operating characteristic curve (AUC) with 95% confidential interval, accuracy, sensitivity, specificity, positive and negative likelihood ratio. Results: Similar mean AUC values were achieved for five models; VGGNet (0.96), ResNet (0.96), DenseNet (0.95), EfficientNet (0.96), and ViT (0.95). The same mean accuracy (0.94) and specificity values (0.96) were yielded for all models. In terms of sensitivity, EffcientNet achieved highest mean value (0.85) while the other models produced slightly lower values range from 0.82 to 0.84. Conclusion: In this study, we demonstrated that DCNNs can classify the staging of liver fibrosis according to METAVIR score with high performance using conventional B-mode images. Among them, EfficientNET that have fewer parameters and computation cost produced highest performance. From the results, we believe that DCNNs based classification of liver fibrosis may allow fast and accurate diagnosis of liver fibrosis without needs of additional equipment for add-on test and may be powerful tool for supporting radiologists in clinical practice.

Super-resolution imaging with moving microbubbles has shown potential in identifying fine details of deep-lying vascular compartments. To image the extravascular targets, the current study has employed nanometer-sized, optically triggered perfluorohexane nanodroplets (PFHnDs). In response to pulsed laser irradiation, the PFHnDs repeatedly vaporize and stochastically recondense, resulting in random changes of ultrasound signals. Our previous study has shown that the stochastic recondensation of the PFHnDs can be used to isolate individual PFHnDs for super-resolution imaging. This study introduces an improved method for super-resolution imaging with ultrafast ultrasound imaging of PFHnDs. The previous method was based on subtraction of two consecutive ultrasound images to detect signals from recondensed, isolated droplets whereas our current method compounds respective multiple pre- and post-recondensation ultrafast ultrasound images prior to subtraction to improve the spatial resolution further. To evaluate the axial and lateral resolutions of our method, we repeatedly imaged a phantom containing PFHnDs using a programmable ultrasound system synchronized with a pulsed laser system. As a result, our method improved the lateral and axial resolutions by 54% and 68%, respectively, over the previous super-resolution imaging approach, indicating that it can be used for localizing extravascular molecular targets with superior accuracy.

With the advances in deep learning, including Convolutional Neural Networks (CNN), automated diagnosis technology using medical images has received considerable attention in medical science. In particular, in the field of ultrasound imaging, CNN trains the features of organs through an amount of image data, so that an expert-level automatic diagnosis is possible only with images of actual patients. However, CNN models are also trained on the features that reflect the inherent bias of the imaging machine used for image acquisition. In other words, when the domain of data used for training is different from that of data applied for an actual diagnosis, it is unclear whether consistent performance can be provided by the domain bias. Therefore, we investigate the effect of domain bias on the model with liver ultrasound imaging data obtained from multiple domains. We have constructed a dataset considering the manufacturer and the year of manufacturing of 8 ultrasound imaging machines. First, training and testing were performed by dividing the entire data, in a commonly used method. Second, we have utilized the training data constructed according to the number of domains for the machine learning process. Then we have measured and compared the performance on internal and external domain data. Through the above experiment, we have analyzed the effect of domains of data on model performance. We show that the performance scores evaluated with the internal domain data and the external domain data do not match. We especially show that the performance measured in the evaluation data including the internal domain was much higher than the performance measured in the evaluation data consisting of the external domain. We also show that 3-level classification performance is slightly improved over 5-level classification by mitigating class imbalance by integrating similar classes. The results highlight the need to develop a new methodology for mitigating the machine bias problem so that the model can work correctly even on external domain data, as opposed to the usual approach of constructing evaluation data in the same domain as the training data.

A power MOSFET linearizer is proposed for a high-voltage power amplifier (HVPA) used in high-frequency pulse-echo instrumentation. The power MOSFET linearizer is composed of a DC bias-controlled series power MOSFET shunt with parallel inductors and capacitors. The proposed scheme is designed to improve the gain deviation characteristics of the HVPA at higher input powers. By controlling the MOSFET bias voltage in the linearizer, the gain reduction into the HVPA was compensated, thereby reducing the echo harmonic distortion components generated by the ultrasonic transducers. In order to verify the performance improvement of the HVPA implementing the power MOSFET linearizer, we measured and found that the gain deviation of the power MOSFET linearizer integrated with HVPA under 10 V DC bias voltage was reduced (−1.8 and −0.96 dB, respectively) compared to that of the HVPA without the power MOSFET linearizer (−2.95 and −3.0 dB, respectively) when 70 and 80 MHz, three-cycle, and 26 dB_m input pulse waveforms are applied, respectively. The input 1-dB compression point (an index of linearity) of the HVPA with power MOSFET linearizer (24.17 and 26.19 dB_m at 70 and 80 MHz, respectively) at 10 V DC bias voltage was increased compared to that of HVPA without the power MOSFET linearizer (22.03 and 22.13 dB_m at 70 and 80 MHz, respectively). To further verify the reduction of the echo harmonic distortion components generated by the ultrasonic transducers, the pulse-echo responses in the pulse-echo instrumentation were compared when using HVPA with and without the power MOSFET linearizer. When three-cycle 26 dB_m input power was applied, the second, third, fourth, and fifth harmonic distortion components of a 75 MHz transducer driven by the HVPA with power MOSFET linearizer (−48.34, −44.21, −48.34, and −46.56 dB, respectively) were lower than that of the HVPA without the power MOSFET linearizer (−45.61, −41.57, −45.01, and −45.51 dB, respectively). When five-cycle 20 dB_m input power was applied, the second, third, fourth, and fifth harmonic distortions of the HVPA with the power MOSFET linearizer (−41.54, −41.80, −48.86, and −46.27 dB, respectively) were also lower than that of the HVPA without the power MOSFET linearizer (−25.85, −43.56, −49.04, and −49.24 dB, respectively). Therefore, we conclude that the power MOSFET linearizer could reduce gain deviation of the HVPA, thus reducing the echo signal harmonic distortions generated by the high-frequency ultrasonic transducers in pulse-echo instrumentation.

The medical ultrasound transducer is a principal component in ultrasound systems, as it significantly influences system performance. The high-voltage power amplifier (HVPA) is the key ultrasound transmitter component and interfaces with the medical ultrasound transducer. Therefore, the performance of the HVPA critically affects the echo signal quality of the ultrasound transducer. As they are inherently non-linear devices, harmonic distortion of echo signals generated by the ultrasound transducers would critically deteriorate the echo signal quality in ultrasound systems. The primary aim of this research is to integrate a novel post-linearizer into the HVPA to suppress harmonic distortion in medical ultrasound transducers. Moreover, this study is based on the assumption that linearizing the HVPA could reduce the harmonic distortion components of the echo signals. The developed HVPA with post-linearizer was tested in an ultrasound microscopy system in order to demonstrate the harmonic suppression capability on the echo signal generated by the ultrasound transducer. When 10 MHz, 5-cycle, 26 dBm input power was sent to the HVPA with and without the post-linearizer circuits, the measured second-, third-, fourth- and fifth- harmonic distortions of a 10 MHz transducer with the post-linearizer (−13.11 dB, −10.81 dB, −16.33 dB, and −22.78 dB) were suppressed to a greater degree than those of the same transducer without the post-linearizer (−4.58 dB, −8.87 dB, −10.51 dB, and −15.31 dB).. Therefore, we conclude that the addition of the post-linearizer to the HVPA is a potentially useful electronic technique for improving echo signal quality in medical ultrasound transducers.

In glucose-stimulated insulin secretion (GSIS) of pancreatic β-cells, the rise of free cytosolic Ca²⁺ concentration through voltage-gated calcium channels (VGCCs) triggers the exocytosis of insulin-containing granules. Recently, mechanically induced insulin secretion pathways were also reported, which utilize free cytosolic Ca²⁺ ions as a direct regulator of exocytosis. In this study, we aimed to investigate intracellular Ca²⁺ responses on the HIT-T15 pancreatic β-cell line upon low-intensity pulsed ultrasound (LIPUS) stimulation and found that ultrasound induces two distinct types of intracellular Ca²⁺ oscillation, fast-irregular and slow-periodic, from otherwise resting cells. Both Ca²⁺ patterns depend on the purinergic signaling activated by the rise of extracellular ATP or ADP concentration upon ultrasound stimulation, which facilitates the release through mechanosensitive hemichannels on the plasma membrane. Further study demonstrated that two subtypes of purinergic receptors, P₂X and P₂Y, are working in a competitive manner depending on the level of glucose in the cell media. The findings can serve as an essential groundwork providing an underlying mechanism for the development of a new therapeutic approach for diabetic conditions with further validation.