Adaptive Spectral Envelope Estimation for Doppler Ultrasound

Automatic Max-Likelihood Envelope Detection Algorithm for Quantitative High-Frame-Rate Ultrasound for Neonatal Brain Monitoring

OBJECTIVE

Post-operative brain injury in neonates may result from disturbed cerebral perfusion, but accurate peri-operative monitoring is lacking. High-frame-rate (HFR) cerebral ultrasound could visualize and quantify flow in all detectable vessels using spectral Doppler; however, automated quantification in small vessels is challenging because of low signal amplitude. We have developed an automatic envelope detection algorithm for HFR pulsed wave spectral Doppler signals, enabling neonatal brain quantitative parameter maps during and after surgery.

METHODS

HFR ultrasound data from high-risk neonatal surgeries were recorded with a custom HFR mode (frame rate = 1000 Hz) on a Zonare ZS3 system. A pulsed wave Doppler spectrogram was calculated for each pixel containing blood flow in the image, and spectral peak velocity was tracked using a max-likelihood estimation algorithm of signal and noise regions in the spectrogram, where the most likely cross-over point marks the blood flow velocity. The resulting peak systolic velocity (PSV), end-diastolic velocity (EDV) and resistivity index (RI) were compared with other detection schemes, manual tracking and RIs from regular pulsed wave Doppler measurements in 10 neonates.

RESULTS

Envelope detection was successful in both high- and low-quality arterial and venous flow spectrograms. Our technique had the lowest root mean square error for EDV, PSV and RI (0.46 cm/s, 0.53 cm/s and 0.15, respectively) when compared with manual tracking. There was good agreement between the clinical pulsed wave Doppler RI and HFR measurement with a mean difference of 0.07.

CONCLUSION

The max-likelihood algorithm is a promising approach to accurate, automated cerebral blood flow monitoring with HFR imaging in neonates.

Resistivity index mapping in Kidney based on ultrasensitive Pulsed-Wave Doppler and automatic spectrogram envelope detection

In recent years, ultrasensitive Pulsed-Wave Doppler (uPWD) ultrasound (US) has emerged as an alternative imaging approach for microcirculation imaging and as a complementary tool to other imaging modalities, such as positron emission tomography (PET). uPWD is based on the acquisition of a large set of highly spatiotemporally coherent frames, which allows high-quality images of a wide field of view to be obtained. In addition, these acquired frames allow calculation of the resistivity index (RI) of the pulsatile flow detected over the entire field of view, which is of great interest to clinicians, for example, in monitoring the transplanted kidney course. This work aims to develop and evaluate a method to automatically obtain an RI map of the kidney based on the uPWD approach. The effect of time gain compensation (TGC) on the visualization of vascularization and aliasing on the blood flow frequency response, was also assessed. A pilot study conducted in patients referred for renal transplant Doppler examination showed that the proposed method provided relative errors of about 15% for RI measurements with respect to conventional pulsed-wave (PW) Doppler.

Stable Automatic Envelope Estimation for Noisy Doppler Ultrasound

Doppler ultrasound technology is widespread in clinical applications and is principally used for blood flow measurements in the heart, arteries, and veins. A commonly extracted parameter is the maximum velocity envelope. However, current methods of extracting it cannot produce stable envelopes in high noise conditions. This can limit clinical and research applications using the technology. In this article, a new method of automatic envelope estimation is presented. The method can handle challenging signals with high levels of noise and variable envelope shapes. Envelopes are extracted from a Doppler spectrogram image generated directly from the Doppler audio signal, making it less device-dependent than existing image-processing methods. The method's performance is assessed using simulated pulsatile flow, a flow phantom, and in vivo ascending aortic flow measurements and is compared with three state-of-the-art methods. The proposed method is the most accurate in noisy conditions, achieving, on average, for phantom data with signal-to-noise ratios (SNRs) below 10 dB, bias and standard deviation of 0.7% and 3.3% lower than the next-best performing method. In addition, a new method for beat segmentation is proposed. When combined, the two proposed methods exhibited the best performance using in vivo data, producing the least number of incorrectly segmented beats and 8.2% more correctly segmented beats than the next best performing method. The ability of the proposed methods to reliably extract timing indices for cardiac cycles across a range of signal quality is of particular significance for research and monitoring applications.

Harnessing Machine Intelligence in Automatic Echocardiogram Analysis: Current Status, Limitations, and Future Directions

Echocardiography (echo) is a critical tool in diagnosing various cardiovascular diseases. Despite its diagnostic and prognostic value, interpretation and analysis of echo images are still widely performed manually by echocardiographers. A plethora of algorithms has been proposed to analyze medical ultrasound data using signal processing and machine learning techniques. These algorithms provided opportunities for developing automated echo analysis and interpretation systems. The automated approach can significantly assist in decreasing the variability and burden associated with manual image measurements. In this paper, we review the state-of-the-art automatic methods for analyzing echocardiography data. Particularly, we comprehensively and systematically review existing methods of four major tasks: echo quality assessment, view classification, boundary segmentation, and disease diagnosis. Our review covers three echo imaging modes, which are B-mode, M-mode, and Doppler. We also discuss the challenges and limitations of current methods and outline the most pressing directions for future research. In summary, this review presents the current status of automatic echo analysis and discusses the challenges that need to be addressed to obtain robust systems suitable for efficient use in clinical settings or point-of-care testing.

Estimation of End-Diastole in Cardiac Spectral Doppler Using Deep Learning

Electrocardiogram (ECG) is often used together with a spectral Doppler ultrasound to separate heart cycles by determining the end-diastole locations. However, the ECG signal is not always recorded. In such cases, the cardiac cycles can be estimated manually from the ultrasound data retrospectively. We present a deep learning-based method for automatic detection of the end-diastoles in spectral Doppler spectrograms. The method uses a combination of a convolutional neural network (CNN) for extracting features and a recurrent neural network (RNN) for modeling temporal relations. In echocardiography, there are three Doppler spectrogram modalities, continuous wave, pulsed wave, and tissue velocity Doppler. Both the training and test data sets include all three modalities. The model was tested on 643 spectrograms coming from different hospitals than in the training data set. For the purposes described in this work, a valid end-diastole detection is defined as a prediction being closer than 60 ms to the reference value. We will refer to these as true detections. Similarly, a prediction farther away is defined as nonvalid or false detections. The method automatically rejects spectrograms where the detection of an end-diastole has low confidence. When setting the algorithm to reject 1.9%, the method achieved 97.7% true detections with a mean error of 14 ms and had 2.5% false detections on the remaining spectrograms.

Adaptive transverse blood velocity estimation in medical ultrasound: A simulation study

Undoubtedly, highly valuable information about vascular anomalies is attained by the examination of the blood flow profile. The chief drawback of the conventional medical ultrasound in preparation of the blood periodogram is the measurement system shortcoming at the beam to flow angles near 90°. Recently, a method based on transverse oscillation (TO) approach, known as "Fourth-order estimation", has been developed to directly estimate the transverse power spectral density (PSD) of the fully transverse blood flow. One of the basic requirements to accomplish acceptable PSDs by this technique is the sufficiently large observation window. In this paper, two adaptive approaches for efficient estimation of the velocity spectrum of a fully transverse flow by a limited observation window length are described. The first proposed adaptive approach is based on the minimum variance adaptive spectral estimation in combination with the well-known TO technique (TO-MV). Then, by exploiting the eigenspace separation of the observed data to eliminate the contribution of the undesired components, the second technique (TO-EIBMV) is developed. The approaches are validated using Field II simulations for pulsating flow. The proposed methods are tested and compared to the conventional TO transverse spectral estimator by metrics of relative standard deviation (RSD) and relative bias (RB). One of the main achievements is the decrement of the required data samples for spectrogram estimation, which leads to a better temporal resolution. Moreover, for the analyzed adaptive techniques, the robustness of the estimation results for the beam to flow angles of 60-90° and vessel depths ranging from 20 mm to 60 mm are investigated.

Adaptive Maximal Blood Flow Velocity Estimation From Transcranial Doppler Echos

OBJECTIVE

Novel applications of transcranial Doppler (TCD) ultrasonography, such as the assessment of cerebral vessel narrowing/occlusion or the non-invasive estimation of intracranial pressure (ICP), require high-quality maximal flow velocity waveforms. However, due to the low signal-to-noise ratio of TCD spectrograms, measuring the maximal flow velocity is challenging. In this work, we propose a calibration-free algorithm for estimating maximal flow velocities from TCD spectrograms and present a pertaining beat-by-beat signal quality index.

METHODS

Our algorithm performs multiple binary segmentations of the TCD spectrogram and then extracts the pertaining envelopes (maximal flow velocity waveforms) via an edge-following step that incorporates physiological constraints. The candidate maximal flow velocity waveform with the highest signal quality index is finally selected.

RESULTS

We evaluated the algorithm on 32 TCD recordings from the middle cerebral and internal carotid arteries in 6 healthy and 12 neurocritical care patients. Compared to manual spectrogram tracings, we obtained a relative error of -1.5%, when considering the whole waveform, and a relative error of -3.3% for the peak systolic velocity.

CONCLUSION

The feedback loop between the signal quality assessment and the binary segmentation yields a robust algorithm for maximal flow velocity estimation. Clinical Impact: The algorithm has already been used in our ICP estimation pipeline. By making the code and the data publicly available, we hope that the algorithm will be a useful building block for the development of novel TCD applications that require high-quality flow velocity waveforms.

Subspace-Based Blood Power Spectral Capon Combined with Wiener Postfilter to Provide a High-Quality Velocity Waveform with Low Mathematical Complexity

In Doppler analysis, the power spectral density (PSD), which accounts for the axial velocity distribution of the blood scatterers, is estimated. The conventional spectral estimator is Welch's method, which suffers from frequency leakage at small observation window length. The performance of adaptive techniques such as blood power Capon (BPC) has been promising at the cost of higher computation complexity. Reducing the computational complexity while retaining the benefits of BPC would be necessary for real-time implementation. The purpose of the work described here was to investigate whether it is possible to decrease the computation load in BPC and still obtain acceptable results. The computation complexity in BPC is owing primarily to the matrix inversion required for computing the PSD estimate. We here propose the subspace blood power Capon technique, which employs a data covariance matrix with reduced number of rows in estimation of the weight vector. In maximum velocity estimation in the spectra, the signal noise slope intersection envelop estimator that makes use of the integrated power spectrum is employed. The evaluations are made based on both simulated and in vivo data. The results indicate that it is possible to reduce the order of complexity to almost 12.25% at the cost of 2.31% and 2.24% increases in the relative standard deviation and relative bias of the estimates. Moreover, the Wiener post-filter as a post-weighting factor, which will be multiplied by the final weight vector of the spectral estimator, estimates the power of the desired signal and the power of the interference plus noise to improve the contrast. The proposed estimator has exhibited a promising performance at beam-to-flow angles of 45°, 60° and 75°. Furthermore, the robust performance of the proposed estimator against variation in the flow rate is also documented.

Automated Spectral Doppler Profile Tracing

OBJECTIVE

The authors hypothesized that automated tracings of both pulsed wave (PW) and continuous wave (CW) Doppler correlate well with manual measurements performed by an experienced echocardiographer.

DESIGN

The authors performed a retrospective analysis of spectral Doppler profile measurements performed by automated software and an echocardiographer.

SETTING

University hospital, single institution.

PARTICIPANTS

The authors reviewed transesophageal echocardiographic examinations from patients undergoing transcatheter aortic valve (AV) replacement procedures at their institution.

INTERVENTIONS

No interventions were performed solely for research purposes.

MEASUREMENTS AND MAIN RESULTS

PW and CW spectral envelopes at the left ventricular outflow tract (LVOT) and AV were analyzed. Blinded, a board-certified echocardiographer performed manual measurements of the identical spectral envelopes. Peak velocities, mean gradients, and velocity time integrals (VTI) were collected. A total of 33 PW as well as 33 CW Doppler spectral envelopes were evaluated. There was no significant difference between the measurements provided by the automated software and manual tracings. LVOT PW VTI automated versus manual: 18.2 cm versus 15.9 cm, p = 0.11. AV CW VTI automated versus manual: 65.8 cm versus 64.8 cm, p = 0.90. AV CW mean gradient automated versus manual: 24.3 mmHg versus 23.4 mmHg, p = 0.84. AV CW peak velocity automated versus manual: 3.00 m/s versus 2.98 m/s, p = 0.93. Correlation coefficients were all above 0.9.

CONCLUSIONS

Automated measurements of peak velocities, mean gradients, and VTI of spectral Doppler correlate closely with manual measurements performed by an experienced echocardiographer.

A Robust Automatic Ultrasound Spectral Envelope Estimation

Accurate estimation of ultrasound Doppler spectrogram envelope is essential for clinical pathological diagnosis of various cardiovascular diseases. However, due to intrinsic spectral broadening in the power spectrum and speckle noise existing in ultrasound images, it is difficult to obtain the accurate maximum velocity. Each of the standard existing methods has their own limitations and does not work well in complicated recordings. This paper proposes a robust automatic spectral envelope estimation method that is more accurate in phantom recordings and various in-vivo recordings than the currently used methods. Comparisons were performed on phantom recordings of the carotid artery with varying noise and additional in-vivo recordings. The accuracy of the proposed method was on average 8% greater than the existing methods. The experimental results demonstrate the wide applicability under different blood conditions and the robustness of the proposed algorithm.

Quantitative Doppler Analysis Using Conventional Color Flow Imaging Acquisitions

Interleaved acquisitions used in conventional triplex mode result in a tradeoff between the frame rate and the quality of velocity estimates. On the other hand, workflow becomes inefficient when the user has to switch between different modes, and measurement variability is increased. This paper investigates the use of power spectral Capon estimator in quantitative Doppler analysis using data acquired with conventional color flow imaging (CFI) schemes. To preserve the number of samples used for velocity estimation, only spatial averaging was utilized, and clutter rejection was performed after spectral estimation. The resulting velocity spectra were evaluated in terms of spectral width using a recently proposed spectral envelope estimator. The spectral envelopes were also used for Doppler index calculations using in vivo and string phantom acquisitions. In vivo results demonstrated that the Capon estimator can provide spectral estimates with sufficient quality for quantitative analysis using packet-based CFI acquisitions. The calculated Doppler indices were similar to the values calculated using spectrograms estimated on a commercial ultrasound scanner.