Ultrasonic Imaging最新文献

Thyroid cancer is one of the common types of cancer worldwide, and Ultrasound (US) imaging is a modality normally used for thyroid cancer diagnostics. The American College of Radiology Thyroid Imaging Reporting and Data System (ACR TIRADS) has been widely adopted to identify and classify US image characteristics for thyroid nodules. This paper presents novel methods for detecting the characteristic descriptors derived from TIRADS. Our methods return descriptions of the nodule margin irregularity, margin smoothness, calcification as well as shape and echogenicity using conventional computer vision and deep learning techniques. We evaluate our methods using datasets of 471 US images of thyroid nodules acquired from US machines of different makes and labeled by multiple radiologists. The proposed methods achieved overall accuracies of 88.00%, 93.18%, and 89.13% in classifying nodule calcification, margin irregularity, and margin smoothness respectively. Further tests with limited data also show a promising overall accuracy of 90.60% for echogenicity and 100.00% for nodule shape. This study provides an automated annotation of thyroid nodule characteristics from 2D ultrasound images. The experimental results showed promising performance of our methods for thyroid nodule analysis. The automatic detection of correct characteristics not only offers supporting evidence for diagnosis, but also generates patient reports rapidly, thereby decreasing the workload of radiologists and enhancing productivity.

The paper presents a novel framework for the prediction of the raised Intracranial Pressure (ICP) from ocular ultrasound images of traumatic patients through automated measurement of Optic Nerve Sheath Diameter (ONSD) and Eyeball Transverse Diameter (ETD). The measurement of ONSD using an ocular ultrasound scan is non-invasive and correlates with the raised ICP. However, the existing studies suggested that the ONSD value alone is insufficient to indicate the ICP condition. Since the ONSD and ETD values may vary among patients belonging to different ethnicity/origins, there is a need for developing an independent global biomarker for predicting raised ICP condition. The proposed work develops an automated framework for the prediction of raised ICP by developing algorithms for the automated measurement of ONSD and ETD values. It is established that the ONSD and ETD ratio (OER) is a potential biomarker for ICP prediction independent of ethnicity and origin. The OER threshold value is determined by performing statistical analysis on the data of 57 trauma patients obtained from the AIIMS, New Delhi. The automated OER is computed and compared with the conventionally measured ICP by determining suitable correlation coefficients. It is found that there is a significant correlation of OER with ICP (r = .81, p ≤ .01), whereas the correlation of ONSD alone with ICP is relatively less (r = .69, p = .004). These correlation values indicate that OER is a better parameter for the prediction of ICP. Further, the threshold value of OER is found to be 0.21 for predicting raised ICP conditions in this study. Scatter plot and Heat map analysis of OER and corresponding ICP reveal that patients with OER ≥ 0.21, have ICP in the range of 17 to 35 mm Hg. In the data available for this research work, OER ranges from 0.17 to 0.35.

Efficient Neural Architecture Search (ENAS) is a recent development in searching for optimal cell structures for Convolutional Neural Network (CNN) design. It has been successfully used in various applications including ultrasound image classification for breast lesions. However, the existing ENAS approach only optimizes cell structures rather than the whole CNN architecture nor its trainable hyperparameters. This paper presents a novel framework for automatic design of CNN architectures by combining strengths of ENAS and Bayesian Optimization in two-folds. Firstly, we use ENAS to search for optimal normal and reduction cells. Secondly, with the optimal cells and a suitable hyperparameter search space, we adopt Bayesian Optimization to find the optimal depth of the network and optimal configuration of the trainable hyperparameters. To test the validity of the proposed framework, a dataset of 1522 breast lesion ultrasound images is used for the searching and modeling. We then evaluate the robustness of the proposed approach by testing the optimized CNN model on three external datasets consisting of 727 benign and 506 malignant lesion images. We further compare the CNN model with the default ENAS-based CNN model, and then with CNN models based on the state-of-the-art architectures. The results (error rate of no more than 20.6% on internal tests and 17.3% on average of external tests) show that the proposed framework generates robust and light CNN models.

Metastases to the thyroid gland (MTT) are uncommon in clinical practice. The ultrasound (US) features are easily confused with primary thyroid malignancy, Hashimoto's thyroiditis, and other thyroid diseases. Therefore, this study aimed to assess the role of US and analysis of prognosis of MTT. A total of 45 patients with MTT in the database between July 2009 and February 2022 at the Fujian Cancer Hospital were reviewed. US examinations were performed only on 20 patients, who were finally included in our study. Among the 20 patients, nine were male, and eleven were female. According to US characteristics, metastases to the thyroid gland were divided into nodular and diffuse types (17 and 3 cases, respectively). Three lesions (17.6%) had circumscribed margins, and 14 (82.4%) were uncircumscribed. Three lesions (17.6%) were regular in shape, and 14 (82.4%) were irregular. Nine metastases (52.9%) were a taller-than-wide shape, and eight (47.1%) were not a taller-than-wide shape. Ten lesions (58.8%) had rich vascularity, and seven (41.2%) had absence/not rich vascularity. The mean overall survival (OS) from the time of MTT diagnosis was 22 months (95% confidence interval: 5.95-38.05). The 1-, 3-, and 5-year OS after metastasis was 68.1%, 25.5%, and 17%, respectively. The prognosis of MTT was poor, which is closely related to the characteristics of the primary tumor and metastatic disease. The US findings and US-guided core needle biopsy may be useful in diagnosing MTT in patients with a history of the malignant tumors.

This biophysical analysis explores the first-principles physics of movement of white blood cell sized particles, suspended in an aqueous fluid and experiencing progressive or standing waves of acoustic pressure. In many current applications the cells are gradually nudged or herded toward the nodes of the standing wave, providing a degree of acoustic focusing and concentration of the cells in layers perpendicular to the direction of sound propagation. Here the underlying biomechanics of this phenomenon are analyzed specifically for the viscous regime of water and for small diameter microscopic spheroids such as living cells. The resulting mathematical model leads to a single algebraic expression for the creep or drift velocity as a function of sound frequency, amplitude, wavelength, fluid viscosity, boundary dimensions, and boundary reflectivity. This expression can be integrated numerically by a simple and fast computer algorithm to demonstrate net movement of particles as a function of time, providing a guide to optimization in a variety of emerging applications of ultrasonic cell focusing.

Methods to assess ultrasound backscatter anisotropy from clinical array transducers have recently been developed. However, they do not provide information about the anisotropy of microstructural features of the specimens. This work develops a simple geometric model, referred to as the secant model, of backscatter coefficient anisotropy. Specifically, we evaluate anisotropy of the frequency dependence of the backscatter coefficient parameterized in terms of effective scatterer size. We assess the model in phantoms with known scattering sources and in a skeletal muscle, a well-known anisotropic tissue. We demonstrate that the secant model can determine the orientation of the anisotropic scatterers, as well as accurately determining effective scatterer sizes, and it may classify isotropic versus anisotropic scatterers. The secant model may find utility in monitoring disease progression as well as characterizing normal tissue architectures.

Accurately estimating all strain components in quasi-static ultrasound elastography is crucial for the full analysis of biological media. In this study, 2D strain tensor imaging was investigated, focusing on the use of a regularization method to improve strain images. This method enforces the tissue property of (quasi-) incompressibility, while penalizing strong field variations, to smooth the displacement fields and reduce the noise in the strain components. The performance of the method was assessed with numerical simulations, phantoms, and in vivo breast tissues. For all the media examined, the results showed a significant improvement in both lateral displacement and strain, while axial fields were only slightly modified by the regularization. The introduction of penalty terms allowed us to obtain shear strain and rotation elastograms where the patterns around the inclusions/lesions were clearly visible. In phantom cases, the findings were consistent with the results obtained from the modeling of the experiments. Finally, the easier detectability of the inclusions/lesions in the final lateral strain images was associated with higher elastographic contrast-to-noise ratios (CNRs), with values in the range of [0.54-9.57] versus [0.08-0.38] before regularization.

Ultrasonography has become an essential part of clinical diagnosis owing to its noninvasive, and real-time nature. To assist diagnosis, automatically segmenting a region of interest (ROI) in ultrasound images is becoming a vital part of computer-aided diagnosis (CAD). However, segmenting ROIs on medical images with relatively low contrast is a challenging task. To better achieve medical ROI segmentation, we propose an efficient module denoted as multiscale attentional convolution (MSAC), utilizing cascaded convolutions and a self-attention approach to concatenate features from various receptive field scales. Then, MSAC-Unet is constructed based on Unet, employing MSAC instead of the standard convolution in each encoder and decoder for segmentation. In this study, two representative types of ultrasound images, one of the thyroid nodules and the other of the brachial plexus nerves, were used to assess the effectiveness of the proposed approach. The best segmentation results from MSAC-Unet were achieved on two thyroid nodule datasets (TND-PUH3 and DDTI) and a brachial plexus nerve dataset (NSD) with Dice coefficients of 0.822, 0.792, and 0.746, respectively. The analysis of segmentation results shows that our MSAC-Unet greatly improves the segmentation accuracy with more reliable ROI edges and boundaries, decreasing the number of erroneously segmented ROIs in ultrasound images.

This study demonstrates the implementation of a shear wave reconstruction algorithm that enables concurrent acoustic radiation force impulse (ARFI) imaging and shear wave elasticity imaging (SWEI) of prostate cancer and zonal anatomy. The combined ARFI/SWEI sequence uses closely spaced push beams across the lateral field of view and simultaneously tracks both on-axis (within the region of excitation) and off-axis (laterally offset from the excitation) after each push beam. Using a large number of push beams across the lateral field of view enables the collection of higher signal-to-noise ratio (SNR) shear wave data to reconstruct the SWEI volume than is typically acquired. The shear wave arrival times were determined with cross-correlation of shear wave velocity signals in two dimensions after 3-D directional filtering to remove reflection artifacts. To combine data from serially interrogated lateral push locations, arrival times from different pushes were aligned by estimating the shear wave propagation time between push locations. Shear wave data acquired in an elasticity lesion phantom and reconstructed using this algorithm demonstrate benefits to contrast-to-noise ratio (CNR) with increased push beam density and 3-D directional filtering. Increasing the push beam spacing from 0.3 to 11.6 mm (typical for commercial SWEI systems) resulted in a 53% decrease in CNR. In human in vivo data, this imaging approach enabled high CNR (1.61-1.86) imaging of histologically-confirmed prostate cancer. The in vivo images had improved spatial resolution and CNR and fewer reflection artifacts as a result of the high push beam density, the high shear wave SNR, the use of multidimensional directional filtering, and the combination of shear wave data from different push beams.