Spatiotemporal Deep Learning-Based Cine Loop Quality Filter for Handheld Point-of-Care Echocardiography.

The reliability of automated image interpretation of point-of-care (POC) echocardiography scans depends on the quality of the acquired ultrasound data. This work reports on the development and validation of spatiotemporal deep learning models to assess the suitability of input ultrasound cine loops collected using a handheld echocardiography device for processing by an automated quantification algorithm (e.g. ejection fraction estimation). POC echocardiograms (n=885 DICOM cine loops from 175 patients) from two sites were collected using a handheld ultrasound device and annotated for image quality at the frame-level. Attributes of high-quality frames for left ventricular (LV) quantification included a temporally-stable LV, reasonable coverage of LV borders, and good contrast between the borders and chamber. Attributes of low-quality frames included temporal instability of the LV and/or imaging artifacts (e.g., lack of contrast, haze, reverberation, acoustic shadowing). Three different neural network architectures were investigated - (a) frame-level convolutional neural network (CNN) which operates on individual echo frames (VectorCNN), (b) single-stream sequence-level CNN which operates on a sequence of echo frames (VectorCNN+LSTM) and (c) two-stream sequence-level CNNs which operate on a sequence of echo and optical flow frames (VectorCNN+LSTM+Average, VectorCNN+LSTM+MinMax, and VectorCNN+LSTM+ConvPool). Evaluation on a sequestered test dataset containing 76 DICOM cine loops with 16,914 frames showed that VectorCNN+LSTM can effectively utilize both spatial and temporal information to regress the quality of an input frame (accuracy: 0.925, sensitivity = 0.860, specificity = 0.952), compared to the frame-level VectorCNN that only utilizes spatial information in that frame (accuracy: 0.903, sensitivity = 0.791, specificity = 0.949). Furthermore, an independent sample t-test indicated that the cine loops classified to be of adequate quality by the VectorCNN+LSTM model had a statistically significant lower bias in the automatically estimated EF (mean bias = - 3.73 ± 7.46 %, versus a clinically obtained reference EF) compared to the loops classified as inadequate (mean bias = -15.92 ± 12.17 %) (p = 0.007). Thus, cine loop stratification using the proposed spatiotemporal CNN model improves the reliability of automated point-of-care echocardiography image interpretation.

In Vivo Adaptive Bayesian Regularized Lagrangian Carotid Strain Imaging for Murine Carotid Arteries and Its Associations With Histological Findings.

OBJECTIVES: Non-invasive methods for monitoring arterial health and identifying early injury to optimize treatment for patients are desirable. The objective of this study was to demonstrate the use of an adaptive Bayesian regularized Lagrangian carotid strain imaging (ABR-LCSI) algorithm for monitoring of atherogenesis in a murine model and examine associations between the ultrasound strain measures and histology. METHODS: Ultrasound radiofrequency (RF) data were acquired from both the right and left common carotid artery (CCA) of 10 (5 male and 5 female) ApoE tm1Unc/J mice at 6, 16 and 24 wk. Lagrangian accumulated axial, lateral and shear strain images and three strain indices-maximum accumulated strain index (MASI), peak mean strain of full region of interest (ROI) index (PMSRI) and strain at peak axial displacement index (SPADI)-were estimated using the ABR-LCSI algorithm. Mice were euthanized (n = 2 at 6 and 16 wk, n = 6 at 24 wk) for histology examination. RESULTS: Sex-specific differences in strain indices of mice at 6, 16 and 24 wk were observed. For male mice, axial PMSRI and SPADI changed significantly from 6 to 24 wk (mean axial PMSRI at 6 wk = 14.10 ± 5.33% and that at 24 wk = -3.03 ± 5.61%, p < 0.001). For female mice, lateral MASI increased significantly from 6 to 24 wk (mean lateral MASI at 6 wk = 10.26 ± 3.13% and that at 24 wk = 16.42 ± 7.15%, p = 0.048). Both cohorts exhibited strong associations with ex vivo histological findings (male mice: correlation between number of elastin fibers and axial PMSRI: rs = 0.83, p = 0.01; female mice: correlation between shear MASI and plaque score: rs = 0.77, p = 0.009). CONCLUSION: The results indicate that ABR-LCSI can be used to measure arterial wall strain in a murine model and that changes in strain are associated with changes in arterial wall structure and plaque formation.

In Vivo Longitudinal Monitoring of Cardiac Remodeling in Murine Ischemia Models With Adaptive Bayesian Regularized Cardiac Strain Imaging: Validation Against Histology.

Adaptive Bayesian regularized cardiac strain imaging (ABR-CSI) uses raw radiofrequency signals to estimate myocardial wall contractility as a surrogate measure of relative tissue elasticity incorporating regularization in the Bayesian sense. We determined the feasibility of using ABR-CSI -derived strain for in vivo longitudinal monitoring of cardiac remodeling in a murine ischemic injury model (myocardial infarction [MI] and ischemia-reperfusion [IR]) and validated the findings against ground truth histology. We randomly stratified 30 BALB/CJ mice (17 females, 13 males, median age = 10 wk) into three surgical groups (MI = 10, IR = 12, sham = 8) and imaged pre-surgery (baseline) and 1, 2, 7 and 14 d post-surgery using a pre-clinical high-frequency ultrasound system (VisualSonics Vevo 2100). We then used ABR-CSI to estimate end-systolic and peak radial (er) and longitudinal (el) strain estimates. ABR-CSI was found to have the ability to serially monitor non-uniform cardiac remodeling associated with murine MI and IR non-invasively through temporal variation of strain estimates post-surgery. Furthermore, radial end-systole (ES) strain images and segmental strain curves exhibited improved discrimination among infarct, border and remote regions around the myocardium compared with longitudinal strain results. For example, the MI group had significantly lower (Friedman's with Bonferroni-Dunn test, p = 0.002) ES er values in the anterior middle (infarcted) region at day 14 (n = 9, 9.23 ± 7.39%) compared with the BL group (n = 9, 44.32 ± 5.49). In contrast, anterior basal (remote region) mean ES er values did not differ significantly (non-significant Friedman's test, χ2 = 8.93, p = 0.06) at day 14 (n = 6, 33.05 ± 6.99%) compared with baseline (n = 6, 34.02 ± 6.75%). Histology slides stained with Masson's trichrome (MT) together with a machine learning model (random forest classifier) were used to derive the ground truth cardiac fibrosis parameter termed histology percentage of myocardial fibrosis (PMF). Both radial and longitudinal strain were found to have strong statistically significant correlations with the PMF parameter. However, radial strain had a higher Spearman's correlation value (εresρ = -0.67, n = 172, p < 0.001) compared with longitudinal strain (εlesρ = -0.60, n = 172, p < 0.001). Overall, the results of this study indicate that ABR-CSI can reliably perform non-invasive detection of infarcted and remote myocardium in small animal studies.

Subaperture Processing-Based Adaptive Beamforming for Photoacoustic Imaging.

Delay-and-sum (DAS) beamformers, when applied to photoacoustic (PA) image reconstruction, produce strong sidelobes due to the absence of transmit focusing. Consequently, DAS PA images are often severely degraded by strong off-axis clutter. For preclinical in vivo cardiac PA imaging, the presence of these noise artifacts hampers the detectability and interpretation of PA signals from the myocardial wall, crucial for studying blood-dominated cardiac pathological information and to complement functional information derived from ultrasound imaging. In this article, we present PA subaperture processing (PSAP), an adaptive beamforming method, to mitigate these image degrading effects. In PSAP, a pair of DAS reconstructed images is formed by splitting the received channel data into two complementary nonoverlapping subapertures. Then, a weighting matrix is derived by analyzing the correlation between subaperture beamformed images and multiplied with the full-aperture DAS PA image to reduce sidelobes and incoherent clutter. We validated PSAP using numerical simulation studies using point target, diffuse inclusion and microvasculature imaging, and in vivo feasibility studies on five healthy murine models. Qualitative and quantitative analysis demonstrate improvements in PAI image quality with PSAP compared to DAS and coherence factor weighted DAS (DAS CF ). PSAP demonstrated improved target detectability with a higher generalized contrast-to-noise (gCNR) ratio in vasculature simulations where PSAP produces 19.61% and 19.53% higher gCNRs than DAS and DAS CF , respectively. Furthermore, PSAP provided higher image contrast quantified using contrast ratio (CR) (e.g., PSAP produces 89.26% and 11.90% higher CR than DAS and DAS CF in vasculature simulations) and improved clutter suppression.

Spatiotemporal Bayesian Regularization for Cardiac Strain Imaging: Simulation and In Vivo Results.

Cardiac strain imaging (CSI) plays a critical role in the detection of myocardial motion abnormalities. Displacement estimation is an important processing step to ensure the accuracy and precision of derived strain tensors. In this paper, we propose and implement Spatiotemporal Bayesian regularization (STBR) algorithms for two-dimensional (2-D) normalized cross-correlation (NCC) based multi-level block matching along with incorporation into a Lagrangian cardiac strain estimation framework. Assuming smooth temporal variation over a short span of time, the proposed STBR algorithm performs displacement estimation using at least four consecutive ultrasound radio-frequency (RF) frames by iteratively regularizing 2-D NCC matrices using information from a local spatiotemporal neighborhood in a Bayesian sense. Two STBR schemes are proposed to construct Bayesian likelihood functions termed as Spatial then Temporal Bayesian (STBR-1) and simultaneous Spatiotemporal Bayesian (STBR-2). Radial and longitudinal strain estimated from a finite-element-analysis (FEA) model of realistic canine myocardial deformation were utilized to quantify strain bias, normalized strain error and total temporal relative error (TTR). Statistical analysis with one-way analysis of variance (ANOVA) showed that all Bayesian regularization methods significantly outperform NCC with lower bias and errors (p < 0.001). However, there was no significant difference among Bayesian methods. For example, mean longitudinal TTR for NCC, SBR, STBR-1 and STBR-2 were 25.41%, 9.27%, 10.38% and 10.13% respectively An in vivo feasibility study using RF data from ten healthy mice hearts were used to compare the elastographic signal-to-noise ratio (SNR e ) calculated using stochastic analysis. STBR-2 had the highest expected SNR e both for radial and longitudinal strain. The mean expected SNR e values for accumulated radial strain for NCC, SBR, STBR-1 and STBR-2 were 5.03, 9.43, 9.42 and 10.58, respectively. Overall results suggest that STBR improves CSI in vivo.

Spatiotemporal Coherence Weighting for In Vivo Cardiac Photoacoustic Image Beamformation.

Photoacoustic (PA) image reconstruction generally utilizes delay-and-sum (DAS) beamforming of received acoustic waves from tissue irradiated with optical illumination. However, nonadaptive DAS reconstructed cardiac PA images exhibit temporally varying noise which causes reduced myocardial PA signal specificity, making image interpretation difficult. Adaptive beamforming algorithms such as minimum variance (MV) with coherence factor (CF) weighting have been previously reported to improve the DAS image quality. In this article, we report on an adaptive beamforming algorithm by extending CF weighting to the temporal domain for preclinical cardiac PA imaging (PAI). The proposed spatiotemporal coherence factor (STCF) considers multiple temporally adjacent image acquisition events during beamforming and cancels out signals with low spatial coherence and temporal coherence, resulting in higher background noise cancellation while preserving the main features of interest (myocardial wall) in the resultant PA images. STCF has been validated using the numerical simulations and in vivo ECG and respiratory-signal-gated cardiac PAI in healthy murine hearts. The numerical simulation results demonstrate that STCF weighting outperforms DAS and MV beamforming with and without CF weighting under different levels of inherent contrast, acoustic attenuation, optical scattering, and signal-to-noise (SNR) of channel data. Performance improvement is attributed to higher sidelobe reduction (at least 5 dB) and SNR improvement (at least 10 dB). Improved myocardial signal specificity and higher signal rejection in the left ventricular chamber and acoustic gel region are observed with STCF in cardiac PAI.

Improving Ultrasound Lateral Strain Estimation Accuracy using Log Compression of Regularized Correlation Function.

Normalized cross-correlation (NCC) function used in ultrasound strain imaging can get corrupted due to signal decorrelation inducing large displacement errors. Bayesian regularization has been applied in an iterative manner to regularize the NCC function and to reduce estimation variance and peak-hopping errors. However, incorrect choice of the number of iterations can lead to over-regularization errors. In this paper, we propose the use of log compression of regularized NCC function to improve subsample estimation. Performance of parabolic interpolation before and after log compression of the regularized NCC function were compared in numerical simulations of uniform and inclusion phantoms. Significant improvement was achieved with the proposed scheme for lateral estimation results. For example, lateral signal-to-noise ratio (SNR) was 10 dB higher after log compression at 3% strain in a uniform phantom. Lateral contrast-to-noise ratio (CNR) was 1.81 dB higher with proposed method at 3% strain in inclusion phantom. No significant difference was observed in axial estimation due to presence of phase information and high sampling frequency. Our results suggest that this simple approach makes Bayesian regularization robust to over-regularization artifacts.

Hierarchical Motion Estimation With Bayesian Regularization in Cardiac Elastography: Simulation and In Vivo Validation.

Cardiac elastography (CE) is an ultrasound-based technique utilizing radio-frequency (RF) signals for assessing global and regional myocardial function. In this work, a complete strain estimation pipeline for incorporating a Bayesian regularization-based hierarchical block-matching algorithm, with Lagrangian motion description and myocardial polar strain estimation is presented. The proposed regularization approach is validated using finite-element analysis (FEA) simulations of a canine cardiac deformation model that is incorporated into an ultrasound simulation program. Interframe displacements are initially estimated using a hierarchical motion estimation framework. Incremental displacements are then accumulated under a Lagrangian description of cardiac motion from end-diastole (ED) to end-systole (ES). In-plane Lagrangian finite strain tensors are then derived from the accumulated displacements. Cartesian to cardiac coordinate transformation is utilized to calculate radial and longitudinal strains for ease of interpretation. Benefits of regularization are demonstrated by comparing the same hierarchical block-matching algorithm with and without regularization. Application of Bayesian regularization in the canine FEA model provided improved ES radial and longitudinal strain estimation with statistically significant ( ) error reduction of 48.88% and 50.16%, respectively. Bayesian regularization also improved the quality of temporal radial and longitudinal strain curves with error reductions of 78.38% and 86.67% ( ), respectively. Qualitative and quantitative improvements were also visualized for in vivo results on a healthy murine model after Bayesian regularization. Radial strain elastographic signal-to-noise ratio (SNRe) increased from 3.83 to 4.76 dB, while longitudinal strain SNRe increased from 2.29 to 4.58 dB with regularization.

Real-Time in Vivo Photoacoustic Imaging in the Assessment of Myocardial Dynamics in Murine Model of Myocardial Ischemia.

Photoacoustic imaging (PAI) is an evolving real-time imaging modality that combines the higher contrast of optical imaging with the higher spatial resolution of ultrasound imaging. We utilized dual-wavelength PAI for the diagnosis and monitoring of myocardial ischemia by assessing variations in blood oxygen saturation estimated in a murine model. The use of high-frequency ultrasound in conjunction with PAI enabled imaging of anatomic and functional changes associated with ischemia. Myocardial ischemia was established in eight mice by ligating the left anterior descending artery (LAD). Longitudinal results reveal that PAI is sensitive to acute myocardial ischemia, with a rapid decline in blood oxygen saturation (p ˂ 0.001) observed after LAD ligation (30 min: 33.05 ± 6.80%, 80 min: 36.59 ± 5.22%, 120 min: 36.70 ± 9.46%, 24 h: 40.55 ± 13.04%) compared with baseline (87.83 ± 5.73%). Variation in blood oxygen saturation was found to be linearly correlated with ejection fraction (%), fractional shortening (%) and stroke volume (µL), with Pearson's correlation coefficient values of 0.66, 0.67 and 0.77, respectively (p ˂ 0.001). Our results indicate that PAI has the potential for real-time diagnosis and monitoring of acute myocardial ischemia.