Detection of Cardiac Quiescence from B-Mode Echocardiography Using a Correlation-Based Frame-to-Frame Deviation Measure

Deep-learning based multibeat echocardiographic cardiac phase detection

BACKGROUND

End-to-end automatic detection of cardiac phase in multibeat echocardiograms is crucial for measuring cardiac parameters in clinical scenarios. However, existing studies face limitations due to the high cost of data annotation and collection, and time-consuming detection processes.

PURPOSE

This study introduces a novel multibeat echocardiographic cardiac phase detection network, EchoPhaseNet, to perform fast and accurate cardiac phase detection of variable-length multibeat echocardiographic sequences, with low annotation costs and limited data.

MATERIALS AND METHODS

Five echocardiographic datasets were used in this study, including a small-scale private dataset, Echo-DT (DrumTower), a medium-scale publicly available dataset, PhaseDetection, and three additional publicly available datasets: EchoNet-Dynamic, CAMUS, and EchoNet-Dynamic-MultiBeat. EchoPhaseNet and four other deep learning-based cardiac phase detection methods were trained and internally validated on the Echo-DT and PhaseDetection datasets (with sample ratios for training, validation, and testing set at 60%:20%:20% and 80%:0%:20%, respectively), and then externally validated on the other three datasets. Model performance was evaluated using GradCAM for qualitative visualization and absolute frame difference (aFD) for quantitative accuracy, with statistical significance assessed using Tukey's test and Benjamini-Hochberg correction, considering corrected p-values < $<$ 0.05 as significant.

RESULTS

The annotation costs and accuracy of end-diastolic (ED) and end-systolic (ES) detection using EchoPhaseNet were compared with those of four other comparison methods. EchoPhaseNet achieves effective specific phase detections using only ED/ES labels, reducing annotation costs and making it applicable to a wider range of detection scenarios compared to all the comparison methods. On the Echo-DT dataset, EchoPhaseNet's mean aFD values for ED and ES detection in the A4C view samples were 1.08 and 1.04, respectively, significantly outperforming three comparison methods in ED detection accuracy (p-values < $<$ 0.01) and comparable to the remaining one (p-values > $>$ 0.05). On the PhaseDetection dataset, EchoPhaseNet's mean aFD values for ED and ES detection were 1.67 and 2.19, respectively, comparable to the detection accuracies of all four comparison methods (p-values > $>$ 0.05). In addition, EchoPhaseNet showed strong generalization ability on multiple external validation datasets. After training on the small-scale Echo-DT dataset, EchoPhaseNet significantly outperformed the four comparison methods (p-values < $<$ 0.01) in ED detection, achieving mean aFD values of 1.67 and 1.11 on the EchoNet-Dynamic and EchoNet-Dynamic-MultiBeat datasets, respectively. After training on the PhaseDetection dataset, EchoPhaseNet significantly outperformed the four compared methods (p-values < $<$ 0.01) in ES detection on the EchoNet-Dynamic dataset, achieving mean aFD value of 2.58. EchoPhaseNet's inference time for a single 32-frame sequence segment is substantially lower than that of the four compared methods, not exceeding 8 ms on an RTX 4080 GPU using the PyTorch deep learning framework.

CONCLUSIONS

EchoPhaseNet exhibits clear advantages over existing studies in data annotation and collection costs, as well as detection speed, and is applicable to a wider range of detection scenarios. It demonstrates good practicality and promising prospects for clinical multibeat echocardiographic cardiac phase detection.

Performance evaluation of computer-aided automated master frame selection techniques for fetal echocardiography

PURPOSE

Fetal echocardiography is widely used for the assessment of fetal heart development and detection of congenital heart disease (CHD). Preliminary examination of the fetal heart involves the four-chamber view which indicates the presence of all the four chambers and its structural symmetry. Examination of various cardiac parameters is generally done using the clinically selected diastole frame. This largely depends on the expertise of the sonographer and is prone to intra- and interobservational errors. To overcome this, automated frame selection technique is proposed for the recognition of fetal cardiac chamber from fetal echocardiography.

METHODS

Three techniques have been proposed in this research study to automate the process of determining the frame referred as "Master Frame" that can be used for the measurement of the cardiac parameters. The first method uses frame similarity measures (FSM) for the determination of the master frame from the given cine loop ultrasonic sequences. FSM makes use of similarity measures such as correlation, structural similarity index (SSIM), peak signal to noise ratio (PSNR), and mean square error (MSE) to identify the cardiac cycle, and all the frames in one cardiac cycle are superimposed to form the master frame. The final master frame is obtained by considering the average of the master frame obtained using each similarity measure. The second method uses averaging of ± 20% from the midframes (AMF). The third method uses averaging of all the frames (AAF) of the cine loop sequence. Both diastole and master frames have been annotated by the clinical experts, and their ground truths are compared for validation. No segmentation techniques have been used to avoid the variability of the performance of various segmentation techniques. All the proposed schemes were evaluated using six fidelity metrics such as Dice coefficient, Jaccard ratio, Hausdorff distance, structural similarity index, mean absolute error, and Pratt figure of merit.

RESULTS

The three proposed techniques were tested on the frames extracted from 95 ultrasound cine loop sequences between 19 and 32 weeks of gestation. The feasibility of the techniques was determined by the computation of fidelity metrics between the master frame derived and the diastole frame chosen by the clinical experts. The FSM-based identified master frame found to closely match with manually chosen diastole frame and also ensures statistically significant. The method also detects automatically the cardiac cycle. The resultant master frame obtained through AMF though found to be identical to that of the diastole frame, the size of the chambers found to be reduced that can lead to inaccurate chamber measurement. The master frame obtained through AAF was not found to be identical to that of clinical diastole frame.

CONCLUSION

It can be concluded that the frame similarity measure (FSM)-based master frame can be introduced in the clinical routine for segmentation followed by cardiac chamber measurements. Such automated master frame selection also overcomes the manual intervention of earlier reported techniques in the literature. The fidelity metrics assessment further confirms the suitability of proposed master frame for automated fetal chamber recognition.

A Thin Transducer With Integrated Acoustic Metamaterial for Cardiac CT Imaging and Gating

Coronary artery disease (CAD) is a leading cause of death globally. Computed tomography coronary angiography (CTCA) is a noninvasive imaging procedure for diagnosis of CAD. However, CTCA requires cardiac gating to ensure that diagnostic-quality images are acquired in all patients. Gating reliability could be improved by utilizing ultrasound (US) to provide a direct measurement of cardiac motion; however, commercially available US transducers are not computed tomography (CT) compatible. To address this challenge, a CT-compatible 2.5-MHz cardiac phased array transducer is developed via modeling, and then, an initial prototype is fabricated and evaluated for acoustic and radiographic performance. This 92-element piezoelectric array transducer is designed with a thin acoustic backing (6.5 mm) to reduce the volume of the radiopaque acoustic backing that typically causes arrays to be incompatible with CT imaging. This thin acoustic backing contains two rows of air-filled, triangular prism-shaped voids that operate as an acoustic diode. The developed transducer has a bandwidth of 50% and a single-element SNR of 9.9 dB compared to 46% and 14.7 dB for a reference array without an acoustic diode. In addition, the acoustic diode reduces the time-averaged reflected acoustic intensity from the back wall of the acoustic backing by 69% compared to an acoustic backing of the same composition and thickness without the acoustic diode. The feasibility of real-time echocardiography using this array is demonstrated in vivo, including the ability to image the position of the interventricular septum, which has been demonstrated to effectively predict cardiac motion for prospective, low radiation CTCA gating.

Ultrasound-gated computed tomography coronary angiography: Development of ultrasound transducers with improved computed tomography compatibility

PURPOSE

Cardiovascular disease (CVD) is a leading cause of death worldwide, with coronary artery disease (CAD) accounting for nearly half of all CVD deaths. The current gold standard for CAD diagnosis is catheter coronary angiography (CCA), an invasive, expensive procedure. Computed tomography coronary angiography (CTCA) represents an attractive non-invasive alternative to CCA, however, CTCA requires gated acquisition of CT data during periods of minimal cardiac motion (quiescent periods) to avoid non-diagnostic scans. Current gating methods either expose patients to high levels of radiation (retrospective gating) or lead to high rates of non-diagnostic scans (prospective gating) due to the challenge of predicting cardiac quiescence based on ECG alone. Alternatively, ultrasound (US) imaging has been demonstrated as an effective indicator of cardiac quiescence, however, ultrasound transducers produce prominent streak artifacts that disrupt CTCA scans. In this study, a proof-of-concept array transducer with improved CT-compatibility was developed for utilization in an integrated US-CTCA system.

METHODS

Alternative materials were tested radiographically and acoustically to replace the radiopaque acoustic backings utilized in low frequency (1-4 MHz) cardiac US transducers. The results of this testing were used to develop alternative acoustic backings consisting of varying concentrations of aluminum oxide in an epoxy matrix via simulations. On the basis of these simulations, single element test transducers designed to operate at 2.5 MHz were fabricated, and the performance of these devices was characterized via acoustic and radiographic testing with micro-computed tomography (micro-CT). Finally, a first proof-of-concept cardiac phased array transducer was developed and its US imaging performance was evaluated. Micro-CT images of the developed US array with improved CT-compatibility were compared with those of a conventional array.

RESULTS

Materials testing with micro-CT identified an acoustic backing with a measured radiopacity of 1008 HU, more than an order of magnitude lower than that of the acoustic backing (24,000 HU) typically used in cardiac transducers operating in the 1-4 MHz range. When utilized in a simulated transducer design, this acoustic backing yielded a -6-dB fractional bandwidth of 57%, similar to the 54% bandwidth of the transducer with the radiopaque acoustic backing. The developed 2.5 MHz, single element transducer based on these simulations exhibited a fractional bandwidth of 51% and signal-to-noise ratio (SNR) of 14.7 dB. Finally, the array transducer developed with the acoustic backing having decreased radiopacity exhibited a 56% fractional bandwidth and 10.4 dB single channel SNR, with penetration depth >10 cm in phantom and in vivo imaging using the full array.

CONCLUSIONS

The first attempt at developing a CT-compatible ultrasound transducer is described. The developed CT-compatible transducer exhibits improved radiographic compatibility relative to conventional cardiac array transducers with similar SNR, bandwidth, and penetration depth for US imaging, according to phantom and in vivo cardiac imaging. A CT-compatible US transducer might be used to identify cardiac quiescence and prospectively gate CTCA acquisition, reducing challenges associated with current gating approaches, specifically relatively high rates of non-diagnostic scans for prospective ECG gating and high radiation dose for retrospective gating.

Multibeat echocardiographic phase detection using deep neural networks

BACKGROUND

Accurate identification of end-diastolic and end-systolic frames in echocardiographic cine loops is important, yet challenging, for human experts. Manual frame selection is subject to uncertainty, affecting crucial clinical measurements, such as myocardial strain. Therefore, the ability to automatically detect frames of interest is highly desirable.

METHODS

We have developed deep neural networks, trained and tested on multi-centre patient data, for the accurate identification of end-diastolic and end-systolic frames in apical four-chamber 2D multibeat cine loop recordings of arbitrary length. Seven experienced cardiologist experts independently labelled the frames of interest, thereby providing infallible annotations, allowing for observer variability measurements.

RESULTS

When compared with the ground-truth, our model shows an average frame difference of -0.09 ± 1.10 and 0.11 ± 1.29 frames for end-diastolic and end-systolic frames, respectively. When applied to patient datasets from a different clinical site, to which the model was blind during its development, average frame differences of -1.34 ± 3.27 and -0.31 ± 3.37 frames were obtained for both frames of interest. All detection errors fall within the range of inter-observer variability: [-0.87, -5.51]±[2.29, 4.26] and [-0.97, -3.46]±[3.67, 4.68] for ED and ES events, respectively.

CONCLUSIONS

The proposed automated model can identify multiple end-systolic and end-diastolic frames in echocardiographic videos of arbitrary length with performance indistinguishable from that of human experts, but with significantly shorter processing time.

Image-Based Methods for Phase Estimation, Gating, and Temporal Superresolution of Cardiac Ultrasound

Objective:

Ultrasound is an effective tool for rapid non-invasive assessment of cardiac structure and function. Determining the cardiorespiratory phases of each frame in the ultrasound video and capturing the cardiac function at a much higher temporal resolution is essential in many applications. Fulfilling these requirements is particularly challenging in preclinical studies involving small animals with high cardiorespiratory rates, requiring cumbersome and expensive specialized hardware.

Methods:

We present a novel method for the retrospective estimation of cardiorespiratory phases directly from the ultrasound videos. It transforms the videos into a univariate time-series preserving the evidence of periodic cardiorespiratory motion, decouples the signatures of cardiorespiratory motion with a trend extraction technique, and estimates the cardiorespiratory phases using a Hilbert transform approach. We also present a robust nonparametric regression technique for respiratory gating and a novel kernel-regression model for reconstructing images at any cardiac phase facilitating temporal super-resolution.

Results:

We validated our methods using 2D echocardiography videos and electrocardiogram (ECG) recordings of 6 mice. Our cardiac phase estimation method provides accurate phase estimates with a mean-phase-error-range of 3–6% against ECG derived phase and outperforms three previously published methods in locating ECGs R-wave peak frames with a mean-frame-error-range of 0.73–1.36. Our kernel-regression model accurately reconstructs images at any cardiac phase with a mean-normalized-correlation-range of 0.81–0.85 over 50 leave-one-out-cross-validation rounds.

Conclusion and Significance:

Our methods can enable tracking of cardiorespiratory phases without additional hardware and reconstruction of respiration-free single cardiac-cycle videos at a much higher temporal resolution.

Automatic detection of end-diastolic and end-systolic frames in 2D echocardiography

BACKGROUND

Correctly selecting the end-diastolic and end-systolic frames on a 2D echocardiogram is important and challenging, for both human experts and automated algorithms. Manual selection is time-consuming and subject to uncertainty, and may affect the results obtained, especially for advanced measurements such as myocardial strain.

METHODS AND RESULTS

We developed and evaluated algorithms which can automatically extract global and regional cardiac velocity, and identify end-diastolic and end-systolic frames. We acquired apical four-chamber 2D echocardiographic video recordings, each at least 10 heartbeats long, acquired twice at frame rates of 52 and 79 frames/s from 19 patients, yielding 38 recordings. Five experienced echocardiographers independently marked end-systolic and end-diastolic frames for the first 10 heartbeats of each recording. The automated algorithm also did this. Using the average of time points identified by five human operators as the reference gold standard, the individual operators had a root mean square difference from that gold standard of 46.5 ms. The algorithm had a root mean square difference from the human gold standard of 40.5 ms (P<.0001). Put another way, the algorithm-identified time point was an outlier in 122/564 heartbeats (21.6%), whereas the average human operator was an outlier in 254/564 heartbeats (45%).

CONCLUSION

An automated algorithm can identify the end-systolic and end-diastolic frames with performance indistinguishable from that of human experts. This saves staff time, which could therefore be invested in assessing more beats, and reduces uncertainty about the reliability of the choice of frame.

Echocardiography as an indication of continuous-time cardiac quiescence

Cardiac computed tomography (CT) angiography using prospective gating requires that data be acquired during intervals of minimal cardiac motion to obtain diagnostic images of the coronary vessels free of motion artifacts. This work is intended to assess B-mode echocardiography as a continuous-time indication of these quiescent periods to determine if echocardiography can be used as a cost-efficient, non-ionizing modality to develop new prospective gating techniques for cardiac CT. These new prospective gating approaches will not be based on echocardiography itself but on CT-compatible modalities derived from the mechanics of the heart (e.g. seismocardiography and impedance cardiography), unlike the current standard electrocardiogram. To this end, echocardiography and retrospectively-gated CT data were obtained from ten patients with varied cardiac conditions. CT reconstructions were made throughout the cardiac cycle. Motion of the interventricular septum (IVS) was calculated from both echocardiography and CT reconstructions using correlation-based, deviation techniques. The IVS was chosen because it (1) is visible in echocardiography images, whereas the coronary vessels generally are not, and (2) has been shown to be a suitable indicator of cardiac quiescence. Quiescent phases were calculated as the minima of IVS motion and CT volumes were reconstructed for these phases. The diagnostic quality of the CT reconstructions from phases calculated from echocardiography and CT data was graded on a four-point Likert scale by a board-certified radiologist fellowship-trained in cardiothoracic radiology. Using a Wilcoxon signed-rank test, no significant difference in the diagnostic quality of the coronary vessels was found between CT volumes reconstructed from echocardiography- and CT-selected phases. Additionally, there was a correlation of 0.956 between the echocardiography- and CT-selected phases. This initial work suggests that B-mode echocardiography can be used as a tool to develop CT-compatible gating techniques based on modalities derived from cardiac mechanics rather than relying on the ECG alone.

Relationship between cardiac quiescent periods derived from seismocardiography and echocardiography

The seismocardiogram (SCG) is a measure of chest wall acceleration due to cardiac motion that could potentially supplement the electrocardiogram (ECG) to more reliably predict cardiac quiescence. Accurate prediction is critical for modalities requiring minimal motion during imaging data acquisition, such as cardiac computed tomography (CT) and magnetic resonance imaging (MRI). For seven healthy subjects, SCG and B-mode echocardiography were used to identify quiescent periods on a beat-by-beat basis. Quiescent periods were detected as time intervals when the magnitude of the velocity signals calculated from SCG and echocardiography were less than a specified threshold. The quiescent periods detected from SCG were compared to those detected from B-mode echocardiography. The quiescent periods of the SCG were found to occur before those detected by echocardiography. A linear relationship between the delay from SCG- to echocardiography-detected phases with respect to heart rate was found. This delay could potentially be used to predict cardiac quiescence from SCG-observed quiescence for use with cardiac imaging modalities such as CT and MRI.

Characterization of cardiac quiescence from retrospective cardiac computed tomography using a correlation-based phase-to-phase deviation measure

PURPOSE

Accurate knowledge of cardiac quiescence is crucial to the performance of many cardiac imaging modalities, including computed tomography coronary angiography (CTCA). To accurately quantify quiescence, a method for detecting the quiescent periods of the heart from retrospective cardiac computed tomography (CT) using a correlation-based, phase-to-phase deviation measure was developed.

METHODS

Retrospective cardiac CT data were obtained from 20 patients (11 male, 9 female, 33-74 yr) and the left main, left anterior descending, left circumflex, right coronary artery (RCA), and interventricular septum (IVS) were segmented for each phase using a semiautomated technique. Cardiac motion of individual coronary vessels as well as the IVS was calculated using phase-to-phase deviation. As an easily identifiable feature, the IVS was analyzed to assess how well it predicts vessel quiescence. Finally, the diagnostic quality of the reconstructed volumes from the quiescent phases determined using the deviation measure from the vessels in aggregate and the IVS was compared to that from quiescent phases calculated by the CT scanner. Three board-certified radiologists, fellowship-trained in cardiothoracic imaging, graded the diagnostic quality of the reconstructions using a Likert response format: 1 = excellent, 2 = good, 3 = adequate, 4 = nondiagnostic.

RESULTS

Systolic and diastolic quiescent periods were identified for each subject from the vessel motion calculated using the phase-to-phase deviation measure. The motion of the IVS was found to be similar to the aggregate vessel (AGG) motion. The diagnostic quality of the coronary vessels for the quiescent phases calculated from the aggregate vessel (PAGG) and IVS (PIV S) deviation signal using the proposed methods was comparable to the quiescent phases calculated by the CT scanner (PCT). The one exception was the RCA, which improved for PAGG for 18 of the 20 subjects when compared to PCT (PCT = 2.48; PAGG = 2.07, p = 0.001).

CONCLUSIONS

A method for quantifying the motion of specific coronary vessels using a correlation-based, phase-to-phase deviation measure was developed and tested on 20 patients receiving cardiac CT exams. The IVS was found to be a suitable predictor of vessel quiescence. The diagnostic quality of the quiescent phases detected by the proposed methods was comparable to those calculated by the CT scanner. The ability to quantify coronary vessel quiescence from the motion of the IVS can be used to develop new CTCA gating techniques and quantify the resulting potential improvement in CTCA image quality.

Detection of quiescent cardiac phases in echocardiography data using nonlinear filtering and boundary detection techniques

We describe an algorithm to detect cardiac quiescence within a heartbeat using nonlinear filtering and boundary detection techniques in echocardiography images. The motivation for detection of these quiescent phases is to provide improved cardiac gating to obtain motion-artifact-free images of the heart at cardiac computed tomography (CT). Currently, cardiac gating is provided through electrocardiography (ECG), which does not provide information about the instantaneous mechanical state of the heart. Our goal is to test if information about the actual mechanical motion of the heart obtained from B-mode echocardiographic data could potentially be used for gating purposes. The nonlinear filtering algorithm presented involves anisotropic diffusion to smoothen the homogeneous regions of the B-mode images while preserving image edges that represent myocardial boundaries. Following this, we detect the boundary of a particular region of interest (ROI) using a thresholding step. The positional changes of this ROI are then observed for quiescent phases over multiple cardiac cycles using the ECG's R-R interval. In a pilot study, seven subjects were imaged in the apical, four-chamber view, and quiescence of the interventricular septum was primarily observed in the diastolic region of the ECG signal. However, the position and length of quiescence vary across multiple heartbeats for the same individual and for different individuals as well. The center of quiescence for the seven patients ranged from 51 to 84 % and did not show a trend with heart rates, which ranged from 54 to 83 beats per minute. The gating intervals based on such analysis of echocardiographic signals could potentially optimize cardiac CT gating.