Photoacoustic imaging (PAI) is a non-ionizing modality that provides high spatiotemporal resolution and resolves optical absorbers at depth, allowing for potential simultaneous therapy monitoring with quantitative molecular imaging. Liposome-encapsulated J-aggregates of indocyanine green (Lipo-JICG) are a PAI agent that can be antibody-targeted and provide high-contrast imaging. Mice with ovarian tumors were injected with either targeted or untargeted Lipo-JICGs and imaged on the MSOT inVision PAI system before injection and immediately, 30-min, and 1-hour post-injection. Lipo-JICG contrast in the tumor was statistically significantly higher in the targeted mice, indicating success in molecular targeting and unmixing accuracy of Lipo-JICGs in vivo.

Abusive head trauma occurs to an estimated 4 out of every 10,000 babies born each year, with a 25% mortality rate. Standard imaging of the trauma involves X-ray CT or MRI, which are not ideal for the necessary repeated imaging of abusive head trauma. Thermoacoustic combined tomography, combining ultrasound tomography, photoacoustic tomography, and radio frequency acoustic tomography is uniquely suited for imaging the skull fractures, brain swelling, subdural hematoma and subdural hygroma associated with abusive head trauma. Thermoacoustic combined tomography is a low cost alternative usable at the bedside for the necessary monitoring and diagnosis of abusive head trauma.

We proposed a wavelet-based denoising method to significantly reduce noise in collected protoacoustic signals and improve the BP identification accuracy with fewer signal averages. The average 1024 signal, which has been used in the published study, was used as the reference to identify the accurate acoustic location. We used Daubechies (db) 4 wavelet transform to decompose the collected signals to recover the useful protoacoustic signals. Our approach was able to identify the BP signal up to the average 8 signals that correspond to the dose (<1.5 Gy). This denoising technique would be useful for future 2D/3D protoacoustic imaging and make protoacoustic clinically useful for proton range verification.

Conventional Doppler ultrasound techniques are limited to the investigation of axial flows as it is not sensitive to the transverse flow component. However, since most of the blood vessels are parallel to the skin surface, the flow is transverse and hence conventional Doppler estimates are not reliable. To tackle this, various methods like cross beam Doppler, cross-correlation methods and multi-transmit multi-receive schemes have been developed in the recent past. Most of the strategies in this regard uses plane wave which are electronically steered at different angles for insonification. In this work, the performance of a triangulation based algorithm with non-steered plane wave transmit is investigated for transverse flows. The algorithm makes a best fit out of the triangulation estimates with different receive angles so that the variability in vector estimation with the receive angles are reduced.

B-mode ultrasound displays hyperechoic and hypoechoic targets as larger and smaller, respectively, compared to the true structure. A method to correct for this distortion would enable B-mode to better represent the true structure. For this work, we investigated training DNN beamformers to reduce this B-mode sizing distortion. Aperture domain DNN beamformers were trained using training data generated from simulated anechoic cysts. The DNN beamformers were trained to suppress signals originating from inside the cyst and to preserve signals originating from outside the cyst. The results suggest that DNN beamformers can be trained to reduce B-mode sizing distortions.

In this work, we present a machine learning method to guide an ultrasound operator towards a selected area of interest. Unlike other automatic medical imaging methods, ultrasound imaging is one of the few imaging modalities where the operator's skill and training are critical in obtaining high quality images. Additionally, due to recent advances in affordability and portability of ultrasound technology, its utilization by non-experts has increased. Thus, there is a growing need for developing intelligent systems that have the ability to assist in both clinical and non-clinical scenarios. We propose a system that leverages machine learning to map real time ultrasound scans to transformation vectors that can guide a user to a target organ or anatomical structure. We present a unique training system that passively collects supervised training data from an expert sonographer and uses this data to train a deep regression network.

Background：The differential diagnosis of benign and malignant thyroid nodules from ultrasound (US) images remained challengeable in clinical practice. We aimed to develop and validate a highly automatic and objective diagnostic model based on the deep learning Radiomics method. Methods：We retrospectively enrolled US images and fine-needle aspiration biopsies from 1645 thyroid nodules. A basic convolutional neural network (CNN) model, a transfer learning model, and a newly designed model named deep learning Radiomics of thyroid (DLRT) (Fig. 1) were used for the investigation. Their diagnostic accuracy was further compared with human observers (one senior and one junior US radiologist). Moreover, the robustness of DLRT over different US instruments was also validated. Conclusions：DLRT shows the best overall performance comparing with other deep learning models and human observers. It holds great promise for improving the differential diagnosis of benign and malignant thyroid nodules.

The paper constructs a new mathematical model based on the Doo-Sabin model. Doo-Sabin gives smooth surfaces suited for medical modelling. The new version allows for sharp vertices and edges, so that it can model a broader variety of shapes. The paper uses the sharp Doo-Sabin to model the right heart ventricle (RV). It then compares the standard Doo-Sabin and the sharp one on modelling the RV of 16 patients, based on ultrasound images of their hearts. A ground truth established by a cardiologist is used to compare them. On average, the sharp Doo-Sabin improved 11.54% on the standard.

High-resolution ultrasonic imaging has been increasingly used in dermatology as a complementary technique for cutaneous lesions assessment. Newly designed portable high-resolution ultrasonic imager operating at 50 MHz provides images with up to 4 mm in depth with axial resolution 50µm. Adaptive signal processing algorithms allow highlighting elements and features of the tissues. Benign melanocytic lesions, skin burns and scars have been characterized. Characterization of skin morphology and estimation the overall skin condition based on elastic modality can potentially be used for diagnostic purposes. The new device is intended to be used in medical and cosmetic dermatology.

A routine 3D transrectal ultrasound (TRUS) volume is usually captured with large slice thickness (e.g. 2-5mm). Such ultrasound images with low out-of-slice resolution affect contouring and needle/seed detection in prostate brachytherapy. The purpose of this study is to develop a deep-learning-based method to construct high-resolution images from routinely captured prostate ultrasound images for brachytherapy.

3D Transmission with reflection ultrasound has been shown effective as a basis for quantitative assessment of breast density that is compatible with existing FDA approved methods. Presently, the use of the reflection image is important in delineation of the breast volume from the water bath. However, the reflection data/image may not be available in some scenarios due to scanner design or equipment malfunction. Furthermore, other data (such as levels of data) may be missing. The Spearman Rank coefficient for correlation of the 3D transmission and reflection ultrasound based quantitative breast density (QBD) was 93% which decreased to 91.5% when reflection image/data were removed. The Spearman r increased again to 95% when smoothing was applied to the speed and attenuation images. The results indicate that even without the reflection data information, the 3D transmission ultrasound characterization of the tissue yields QBD values commensurate with FDA approved methods.

In this work we present the development of a statistical shape model (SSM) generated from a set of 13 endocardial landmarks extracted from tri-plane 2D US images and its application to rapidly and accurately estimate the LV blood pool volumes and LVEF of new patient images using multi-view 2D ultrasound. By investigating the use of the SSM with and without size normalization, we show the estimated LVEFs were within less than 2.9% and 1.1% of the ground truth values, respectively. Furthermore, the first 5 most dominant modes of variation described nearly 87% of the overall variance in the SSM.

We present a refraction-corrected sound speed reconstruction technique for layered media based on the angular coherence of plane waves. Previous work have utilized the coherence factor and full-synthetic aperture channel data to successfully estimate sound speed. However, methods for acquiring full-synthetic aperture data require a large number of transmissions, which can confound sound speed estimation due to the scatterer motion between transmit events. By replacing spatial coherence with angular coherence and applying a similar sound-speed estimation methodology, we significantly reduce acquisition time and the computational cost of sound speed estimation while maintaining the accuracy of our sound speed estimates.

Breast cancer screening with mammography and ultrasonography is expected to improve sensitivity compared with mammography alone. An automated breast volume scanner (ABVS) provides the operator-independent whole breast data. However, large volumetric data in screening practice increase radiologists’ workload. Therefore, our goal is to develop an automated lesion detection scheme in ABVS data. In our previous study, we proposed a mass detection scheme using convolutional neural network. One short coming is that the number of cases with lesions is small. In this study, we investigated the automatic mass detection scheme which does not require labeled data.

We present transmission ultrasound as a method to visualize and differentiate fibroglandular tissue within the breast and use an unsupervised learning-based method to quantitatively assess the respective breast density. The method includes initial separation of the breast from the surrounding water bath followed by segmentation of fibroglandular tissue from fat using fuzzy C-means classification. We apply these methods to both tissue phantoms and clinical breast images. The results demonstrate a strong correlation with both BI-RADS and Volpara assessed breast density estimates and imply that breast density as assessed by transmission ultrasound can be of significant clinical utility.

3D ultrasound imaging can be used in the context of robotic needle steering to reach a physical target with a flexible, steerable needle. During the insertion, the tissue may be deformed, thus requiring to track intra-operatively the target’s displacement. This paper proposes a new approach for computationally inexpensive and precise tracking of a moving target in 3D B-mode ultrasound volumes. It interconnects intensity based tracking and motion estimation algorithms. An experimental validation on ex-vivo tissue is proposed with promising tracking precision (estimated average error of 0.3mm) while significantly lowering the computational cost when compared to classical block matching based tracking.

Image guidance aids neurosurgeons in making critical clinical decisions of safe maximal resection of diseased tissue. The brain however undergoes non-linear structural deformation on account of dura opening and tumor resection. Deformable registration of pre-operative ultrasound to intra-operative ultrasound may be used in mapping of pre-operative planning MRI to intraoperative ultrasound. In this work, deformable registration of segmented brain structures in pre- and intraoperative 3D ultrasound was used to register pre- and intraoperative ultrasound. We achieve a mean target registration error of 1.43±0.8 mm when validated with 17 patients pre- and intraoperative ultrasound images of a public dataset.

In this paper, the goal was to develop a reliable and repeatable process to create realistic, kidney phantoms with tunable image visibility under ultrasound and CT imaging modalities for renal biopsy evaluation. The final complex kidney phantom was devised containing kidney cortex, medulla, and ureter. Simultaneously, some lesions were integrated into the phantom to mimic the presence of tumors during a biopsy. The phantoms were created and scanned by ultrasound and CT scanners and the result was a successful advancement in knowledge of materials with acoustic and impedance properties to replicate human organs for the field of image-guided interventions.

Gynecologic interstitial brachytherapy requires multiple needles to be precisely inserted into the tumor to deliver optimal treatment and avoid nearby healthy organs. We propose a 360° 3D transvaginal ultrasound (TVUS) guidance system for visualization of needles and report on the implementation of two automatic needle segmentation algorithms to aid the localization of needles intraoperatively, enhancing the clinical utility and translation of this system. A convolutional neural network approach and a randomized 3D Hough transform-based method were developed and tested for intraoperative 2D and 3D TVUS images, respectively, allowing for immediate adjustments to needle trajectories and potentially improving implant quality.

We proposed a novel data acquisition method with a ring-arrayed transducer to enable forward-viewing US imaging for being compatible with the guidance of needle insertion. The workflow of ultrasound (US)-guided biopsy requires a physician to hold the US probe to locate the interesting area with the non-dominant hand because the dominant hand is occupied by the biopsy needle. Although 3D US imaging is an emerging alternative that provides geometrical information better and used in the needle intervention, it is not enough to perform the low-cost imaging and the co-registered guidance for the needle intervention. Therefore, there is a demand for a solution providing simple/intuitive and accurate guidance and localization of the needle to elevate the conventional US-guided biopsy.

We designed a comprehensive evaluation method including various contrast conditions using a single oil-gel-based phantom by changing the liquid (water or salt water) and temperature (15, 17.5, 20, 22.5, 25, 27.5, and 30°C). The filtered back projection (FBP) and full waveform inversion (FWI) were evaluated. The mean error of the speeds of sound in inclusions with the FBP and FWI were 17.1 ± 14.9 and 8.8 ± 10.1 [m/s], respectively. The mean percentage error of the sizes of the phantom and inclusions with the FBP and FWI were 22.5 ± 22.5% and 3.9 ± 4.3%, respectively.

We are performing clinical studies on breast cancer examination in Hokkaido University Hospital with an ultrasound computed tomography (USCT) system. Through these studies, some reflection images showed the intensity inhomogeneity because ultrasound wave go non-vertically into object surface. This trend increases the burden of interpretation. Therefore, we developed a calibration method to remove this heterogeneity based on the distribution of incident angle of wave which are estimated from the slope of subject surface morphologically extracted from multi-slice reflection images. By applying this correction method to clinical images, the image contrast and uniformity were successfully recovered.

In this paper, a preprocessing method, variational mode decomposition, is used to improve the ultrasound computed tomography image quality. The RF signals are decomposed in the frequency domain using VMD method so that the noise components can be removed. The ultrasound computed tomography image quality is enhanced using the remaining components. Validation is performed through measurement on the breast phantom. Experimental results show that the preprocessing method improved the image resolution and contrast and reduced the noises.

Quickly developed medical market of high-resolution imaging requires suitable phantoms simulating characteristics of skin pathologies to develop new imaging systems and image processing algorithms. The purpose of this study was to create multilayered phantoms mimicking mechanical and acoustic properties of healthy skin and melanoma lesions for high-frequency ultrasonic imaging and quantitative assessment. Phantoms with the properties similar to human skin have been fabricated. Ultrasonic testing of phantoms was performed using scanning acoustic impedance microscope (Honda 50-SI) and hand-held skin imager. The magnitude of the sound speed and attenuation coefficient was measured to be close to those in the human tissue.

Intense pulsed light (IPL) is a high-intensity treatment for skin disorders and ageing. Poorly regulated and inadequately studied, we investigate IPL as a cosmetic device and its effects on collagen dermal components of skin and lab models. Ex-vivo porcine skin models were irradiated with an IPL (584 nm) at 40 J/cm2 and then characterized using AFM and PLM and US. Differences in the structure were observed between the control ex vivo tissue when compared with heavily irradiated tissue at higher IPL intensities. Ex-vivo porcine tissue demonstrated loss of D-banding and gelatinization with increasing dermal depth.

Current molecular imaging modalities face barriers to clinical implementation, providing a need for an improved clinical molecular imaging approach. We hypothesize that EGFR-targeted perfluorocarbon nanodroplets can label metastatic cells; they can then be activated (i.e., converted to a microbubble) and imaged using ultrasound in order to provide a molecular contrast agent that can inform treatment. Pulse sequences were developed for the Verasonics Vantage 128 system to activate and image dodecafluoropentane and dodecafluorohexane nanodroplets. Dodecafluoropentane nanodroplets provided 28-dB enhancement when imaged with pulse-inversion US in a tissue-mimicking environment, while dodecafluorohexane nanodroplets showed activation and subsequent recondensation, allowing for super-resolution imaging.

We are developing a methodology for 3D assessment of ciliary body of the eye, an important but understudied tissue behind the iris opaque to optical imaging, using our new 3D ultrasound biomicroscopy (3D-UBM) imaging system. The ciliary body is important in clinical studies of glaucoma and focusing accommodation. In this preliminary study, using 3D-UBM images, we created unique en face visualizations of the ciliary processes and measured volumes and surface areas, among other biometrics, using a deep-learning semi-automated approach. This work will propel research in ciliary body assessment in clinical diagnosis of glaucoma, treatment planning, and treatment assessment.

Ultrasonic imaging is a non-invasive technique capable of providing high resolution qualitative and quantitative information in real-time. This, coupled with new algorithms has allowed for the extraction and evaluation of human skin tissue for medical and cosmetic applications. Here we investigate and compare its functions to that of atomic force microscopy (AFM) on young vs old epidermal and dermal collagen.

Skin thickness measurements were obtained using two high frequency ultrasound techniques (LOGIQe (GE) and Dermascan) on 37 healthy male human research volunteers (soldiers - median age 20.3 ± 2.9 years) to quantify skin properties and provide a comprehensive and biologically representative data set of skin thicknesses (epidermis and dermis) to be used in developing skin models for the head and hand and updating skin models (if needed) used for predicting burn injury in clothing systems. The study focused on the dominant hand however dominant and non-dominant hand scans were performed on 17 subjects to determine if differences existed.

Rapid brain imaging is central to the diagnosis and treatment of acute neurological conditions, but existing imaging methods require large, immobile, high-power instruments that are nearly impossible to deploy outside specialized environments. Here, we present an alternative to current established neuroimaging modalities, magnetic resonance imaging and X-ray computed tomography, based on ultrasound full-waveform inversion (FWI). We demonstrate the potential of our method in differential stroke diagnosis, using an in-silico model, and show that FWI is capable of reconstructing the location, size, shape and acoustic properties of the pathology.

The ability to conduct ‘partial field of view’ scans in QT ultrasound 3D transmission ultrasound is investigated. Three scenarios are investigated: breast, whole body, and orthopedic imaging. 180 views at 2 degree intervals and 192 mm in the vertical direction were collected from the QT ultrasound™ scanner. Subsequent reconstructions were carried out with a contiguous sector of missing angles varying from 8, 16, 32, 40, 60 and 92 degrees. The difference between the corresponding images was quantitatively and qualitatively analyzed and compared. It was found that a lacuna of 32 degrees did not significantly (in the clinical sense) degrade image quality. Data sparsification in the vertical direction on allowable lacuna size was also addressed. The open acquisition scenario allows us to carry out medical intervention as well as potentially decrease patient anxiety. The importance of 3D data and clinical implications are discussed.

We establish the quantitative accuracy of 3D transmission ultrasound imaging using both fresh and cadaver tissue. The inverse scattering formulation gives high resolution quantitative results even in proximity to bone. The use of fresh tissue obviates the objection that cadaverous tissue has different speed of sound (SOS) due to fixation. We report that fixation changes the SOS by 0.4% or less in fibroglandular tissue, fat and skin. The use of multiple transverse sections at varying distances from the tibio-femoral space ameliorates bias ameliorates bias. We use a 6 mm diameter ROI at 20 successive transverse levels to measure the SOS of Vastus medialis at varying distances from the tibio-femoral space and correlate to accepted literature values (1573 m/s vs 1588.4 m/s). Similar results are shown for fat (1438.7 vs 1440 m/s) and cartilage (1655.4 vs 1660 m/s). We map out regions that have no MR response, but are quantitatively imaged in 3D transmission ultrasound

Diagnostic ultrasound is ubiquitous in clinical practice because it is safe, portable, inexpensive, has high spatial resolution and is real time. Therefore, improving the capabilities of diagnostic ultrasound is a highly significant clinically. In this talk we will discuss different applications of quantitative ultrasound (QUS) imaging and how QUS approaches have evolved over time. Specifically, we will discuss the use of spectral-based approaches to estimate the backscatter coefficient (BSC) and attenuation slope and the use of envelope statistics to describe underlying tissue microstructure. These QUS approaches have been successful at classifying tissue state, monitoring focused ultrasound therapy, detecting early response of breast cancer to neoadjuvant chemotherapy and the automatic detection of nerves in the imaging field. We will demonstrate how QUS approaches can be incorporated on breast tomography machines, which allow an expansion of the tradeoff between spatial resolution and the variance of QUS estimates. One of the ongoing issues with QUS is the inability to properly account for losses in tissues that affect the estimates of the backscatter coefficient. We will demonstrate new calibration procedures that can improve the ability to account for tissue losses. Finally, we will discuss how machine learning approaches can further improve QUS techniques by eliminating the need for models and in some cases eliminating the need for a reference scan.

Mixed Reality (MR) headsets have the potential to revolutionize the way we work, learn, communicate, and get entertained. The main pillars for MR development are wearable, visual and social comfort, as well as immersion experience. Do these pillars intersect the specific needs of medical imaging and surgical guidance applications? We will review the various challenges to implement MR hardware specifically adapted to such tasks with today’s start of the art MR headset technology.

Objective:We aimed to evaluate the performance of the newly developed deep learning Radiomics of elastography (DLRE) for assessing liver fibrosis stages. DLRE adopts the radiomic strategy for quantitative analysis of the heterogeneity in two-dimensional shear wave elastography (2D-SWE) images. Design :A study was conducted to assess its accuracy in chronic hepatitis B (CHB) patients, in comparison with 2D-SWE, transaminase-to-platelet ratio index (APRI), and fibrosis index based on the four factors (FIB-4), by using liver biopsy as the reference standard. Its accuracy and robustness were also investigated by applying different number of acquisitions and different training cohorts, respectively. Conclusion :DLRE shows the best overall performance in predicting liver fibrosis stages comparing with 2D-SWE and biomarkers. It is valuable and practical for the noninvasive accurate diagnosis of liver fibrosis stages in HBV infected patients.

In this study, we propose an approach based on a weekly supervised method for MR-transrectal ultrasound (TRUS) image registration.

Synthetic Aperture Focusing Technique (SAFT) for reflectivity ultrasound tomography is often used in medical applications. SAFT has limitations due the low signal-to-noise ratio. To contribute to a solution, we propose a improvement called weighted-SAFT. It were used k-wave toolbox for simulation in numerical phantoms, Fast Marching Method to phase aberration correction and time of flight to calculated the SAFT weighting. The SAFT reconstructions were made using the raw received signal and, separately, the envelope signal. It is observed how the weighting SAFT leads to a more balanced distribution of intensity values throughout the image, improving its contrast. SAFT-reconstructions using the signal envelope were not good.

Breast cancer detection using Ultrasound computed tomography (USCT) usually works with a collection of ultrasound scans that measure the pressure wavefield emitted by individual transducers. This often requires a large number of transducer pairs to obtain a good coverage of the domain of interest, and careful calibration of the transducers using reference measurements in water. We present a novel approach to obtain time-of-flight measurements between transducer pairs in an USCT aperture by applying the interferometry principle. By substituting active emitters with virtual ones, specific source imprints are eliminated, thus avoiding the need for reference measurements and calibration. Using interferometry and virtual emitters, we retrieve Green’s functions between any two measurement locations, which can be used as new data for the inverse problem. The proposed method gives new perspectives to shorten the acquisition time of an entire USCT data set and to increase the data coverage.

Photoacoustic imaging is a hybrid imaging modality that has the advantages of the high contrast in optical imaging and the high resolution in ultrasound imaging. Quantitative photoacoustic imaging uses one or several optical wavelengths to quantify the concentration of an optical absorber which is needed for molecular and functional imaging. A medical image should, ideally, present the spatial distribution of an intrinsic property of the imaged tissue, however, the conventional photoacoustic images show the spatial distribution of the absorbed energy which depends on both the tissue absorption coefficient and the optical fluence distribution. Herein, we propose a simple improvement to photoacoustic imaging by obtaining a corrected photoacoustic image that depends only on tissue property. This is achieved by non-iteratively solving the inverse problem of light diffusion in tissue assuming that the conventional (uncorrected image) is proportional to the absorbed energy.

In this paper, we propose a workflow for multi-needle detection in 3D ultrasound (US) images with corresponding CT images used for weak supervision. Since the CT images do not exactly match US images, we propose a novel sparse model, dubbed Bidirectional Convolutional Sparse Coding (BiCSC), to tackle this weakly supervised problem. BiCSC aims to extract the latent features from US and CT and then build the relationship between US and CT, where the learned features from the US yield to the features from CT.

Accurate motion tracking of left ventricle (LV) in echocardiography is crucial for locating the abnormal wall motion movements in the heart post-injury. However, in vivo datasets, like patient data and animal studies, lack ground-truth for efficient training of deep learning algorithms. In this work, we propose an unsupervised deformable registration framework that does not require ground-truth data. We demonstrate the tracking of mid-cavity slice of LV and regularize the displacements by enforcing physiological constraints like divergence free and segmentation term. Results on the in vivo canine echocardiography dataset show promising results compared to traditional motion tracking algorithm, specifically non-rigid registration.

In this work we develop and validate a deep fully convolutional neural network architecture for the multi-structure segmentation of echocardiography, including the left ventricular endocardium and epicardium, and the left atrium. We use the recently released CAMUS echocardiography dataset. In ten-fold cross validation with data augmentation, we obtain mean Dice overlaps of 0.93, 0.95, and 0.89 on the three structures respectively, representing state of the art on this dataset. We also report agreement with reported clinical indices, including median absolute errors for left ventricular diastolic (7.4mL) and systolic volumes (4.8mL), and ejection fraction (4.1%), within previously reported inter-observer variability. These encouraging results must still be validated against large-scale independent clinical data.

Reducing speckle noise is an important task for improving visual and automated assessment of retinal OCT images. Traditional image/signal processing methods only offer moderate speckle reduction; deep learning methods can be more effective but require substantial training data, which may not be readily available. We present a novel self-fusion method that offers effective speckle reduction comparable to deep learning methods, but without any external training data. We present qualitative and quantitative results in a variety of datasets from fovea and optic nerve head regions, with varying SNR values for input images.