Conventional medical imaging technologies, including ultrasound, have continued to improve over the years. For example, in oncology, medical imaging is characterized by high sensitivity, i.e., the ability to detect anomalous tissue features, but the ability to classify these tissue features from images often lacks specificity. As a result, a large number of biopsies of tissues with suspicious image findings are performed each year with a vast majority of these biopsies resulting in a negative finding. To improve specificity of cancer imaging, quantitative imaging techniques can play an important role. Conventional ultrasound B-mode imaging is mainly qualitative in nature. However, quantitative ultrasound (QUS) imaging can provide specific numbers related to tissue features that can increase the specificity of image findings leading to improvements in diagnostic ultrasound. QUS imaging can encompass a wide variety of techniques including spectral-based parameterization, elastography, shear wave imaging, flow estimation, and envelope statistics. Currently, spectral-based parameterization and envelope statistics are not available on most conventional clinical ultrasound machines. However, in recent years, QUS techniques involving spectral-based parameterization and envelope statistics have demonstrated success in many applications, providing additional diagnostic capabilities. Spectral-based techniques include the estimation of the backscatter coefficient (BSC), estimation of attenuation, and estimation of scatterer properties such as the correlation length associated with an effective scatterer diameter (ESD) and the effective acoustic concentration (EAC) of scatterers. Envelope statistics include the estimation of the number density of scatterers and quantification of coherent to incoherent signals produced from the tissue. Challenges for clinical application include correctly accounting for attenuation effects and transmission losses and implementation of QUS on clinical devices. Successful clinical and preclinical applications demonstrating the ability of QUS to improve medical diagnostics include characterization of the myocardium during the cardiac cycle, cancer detection, classification of solid tumors and lymph nodes, detection and quantification of fatty liver disease, and monitoring and assessment of therapy.

In this study, the speed of sound (SOS) of two types of rat livers (eight normal livers, four cirrhotic livers) was measured with a scanning acoustic microscope using two transducers, one of which had an 80-MHz and the other a 250-MHz center frequency. The 250-MHz transducer had a better spatial resolution adapted to studying fiber or hepatic parenchymal cells. In normal livers, averages of the SOS values were from 1598 to 1677 m/s at 80-MHz and from 1568 to 1668 m/s at 250-MHz. In the fiber tissue of cirrhotic livers, averages of the SOS values were from 1645 to 1658 m/s at 80-MHz and from 1610 to 1695 m/s at 250-MHz, while the SOS values in the other tissue of cirrhotic livers ranged from 1644 to 1709 m/s at 80-MHz and from 1641 to 1715 m/s at 250-MHz. In one liver, SOS in fiber tissue was larger than that of tissues without fiber while in others it was lower. The resulting two-dimensional SOS maps provide a unique quantitative insight of liver acoustic microstructures in a healthy liver and in a cirrhotic ones. This study would be helpful to understand the complex relationship between acoustic properties and liver disease including fiber tissue.

Ex vivo assessment of microscale tissue biomechanical properties of the mammalian retina could offer insights into diseases such as keratoconus, and macular degeneration. A 250-MHz scanning acoustic microscope (7-μm resolution) has been constructed to derive two-dimensional quantitative maps of attenuation (α), speed of sound (c), acoustic impedance (Z), bulk modulus (B), and mass density (ρ). The two-dimensional maps were compared to coregistered hematoxylin-and-eosin stained sections. This study is the first to quantitatively assess α, c, Z, B, and ρ of individual retinal layers of mammalian animals at high ultrasound frequencies. Significant differences in these parameters between the layers were demonstrated.

Quantitative acoustic microscopy (QAM) at frequencies exceeding 100 MHz has become an established imaging tool to depict acoustical and mechanical properties of soft biological tissues at microscopic resolutions. In this study we investigate a novel autoregressive (AR) model to improve signal processing and parameter estimation and to test its applicability to QAM. The performance of the AR model for estimating acoustical parameters of soft tissues (i.e., acoustic impedance, speed of sound, and attenuation) was compared to the performance of the Hozumi model using simulated ultrasonic QAM-signals and using experimentally measured signals from thin (i.e., 12 and 6 μm) sections of human lymph-node and pig-cornea tissue specimens. Results showed that the AR and Hozumi methods performed equally well (i.e., produced an estimation error of 0) in signals with low, linear attenuation in the tissue and high impedance contrast between the tissue and the coupling medium. However, the AR model outperformed the Hozumi model in estimation accuracy and stability (i.e., parameter error variation and number of outliers) in cases of (1) thin tissuesample thickness and high tissue-sample speed of sound, (2) small impedance contrast between the tissue sample and the coupling medium, (3) high attenuation in the tissue sample and (4) nonlinear attenuation in the tissue sample. Furthermore, the ARmodel allows estimating the exponent of non-linear attenuation. The results of this study suggest that the AR-model approach can improve current QAM by providing more-reliable, quantitative, tissue-property estimates, and also provides additional values of parameters related to non-linear attenuation.

Segmentation and mutant classification of high-frequency ultrasound (HFU) mouse embryo brain ventricle (BV) and body images can provide valuable information for developmental biologists. However, manual segmentation and identification of BV and body requires substantial time and expertise. This article proposes an accurate, efficient and explainable deep learning pipeline for automatic segmentation and classification of the BV and body. For segmentation, a two-stage framework is implemented. The first stage produces a low-resolution segmentation map, which is then used to crop a region of interest (ROI) around the target object and serve as the probability map of the autocontext input for the second-stage fine-resolution refinement network. The segmentation then becomes tractable on high-resolution 3-D images without time-consuming sliding windows. The proposed segmentation method significantly reduces inference time (102.36-0.09 s/volume ≈ 1000× faster) while maintaining high accuracy comparable to previous sliding-window approaches. Based on the BV and body segmentation map, a volumetric convolutional neural network (CNN) is trained to perform a mutant classification task. Through backpropagating the gradients of the predictions to the input BV and body segmentation map, the trained classifier is found to largely focus on the region where the Engrailed-1 (En1) mutation phenotype is known to manifest itself. This suggests that gradient backpropagation of deep learning classifiers may provide a powerful tool for automatically detecting unknown phenotypes associated with a known genetic mutation.