Characterization of the Korotkoff sounds using joint time-frequency analysis

The fundamental mechanisms of the Korotkoff sounds generation

Blood pressure measurement is the most widely performed clinical exam to predict mortality risk. The gold standard for its noninvasive assessment is the auscultatory method, which relies on listening to the so-called "Korotkoff sounds" in a stethoscope placed at the outlet of a pneumatic arm cuff. However, more than a century after their discovery, the origin of these sounds is still debated, which implies a number of clinical limitations. We imaged the Korotkoff sound generation in vivo at thousands of images per second using ultrafast ultrasound. We showed with both experience and theory that Korotkoff sounds are paradoxically not sound waves emerging from the brachial artery but rather shear vibrations conveyed in surrounding tissues by the nonlinear pulse wave propagation. When these shear vibrations reached the stethoscope, they were synchronous, correlated, and comparable in intensity with the Korotkoff sounds. Understanding this mechanism could ultimately improve blood pressure measurement and provide additional understanding of arterial mechanical properties.

B3X: a novel efficient algorithm for accurate automated auscultatory blood pressure estimation

Objective.The auscultatory technique is still considered the most accurate method for non-invasive blood pressure (NIBP) measurement, although its reliability depends on operator's skills. Various methods for automated Korotkoff sounds analysis have been proposed for reliable estimation of systolic (SBP) and diastolic (DBP) blood pressures. To this aim, very complex methodologies have been presented, including some based on artificial intelligence (AI). This study proposes a relatively simple methodology, named B3X, to estimate SBP and DBP by processing Korotkoff sounds recordings acquired during an auscultatory NIBP measurement.Approach.The beat-by-beat change in morphology of adjacent Korotkoff sounds is evaluated via their cross-correlation. The time series of the beat-by-beat cross-correlation and its first derivative are analyzed to locate the timings of SBP and DBP values. Extensive tests were performed on a public database of 350 annotated measurements, and the performance was evaluated according to the BHS, AAMI/ANSI, and International Organization for Standardization (ISO) quality standards.Main results.The proposed approach achieved 'A' scores for SBP and DBP in the BHS grading system, and passed the quality tests of AAMI/ANSI and ISO standards. The B3X algorithm outperformed two well-established algorithms for oscillometric NIBP measurement in both SBP and DBP estimation. It also outperformed four AI-based algorithms in DBP estimation, while providing comparable performance for SBP, at the cost of a much lower computational burden. The full code of the B3X algorithm is provided in a public repository.Significance.The very good performances ensured by the proposed B3X algorithm, at a low computational cost and without the need for parameter training, support its direct implementation into clinical blood pressure (BP) monitoring devices. The results of this study pave the way for solving/overcoming the trade-off between the accuracy of the auscultatory technique and the objectivity of oscillatory measurements, by bringing an automated auscultatory BP measurement method in clinical practice.

Masking of Korotkoff sounds used in blood pressure measurement through auscultation

Measurement of blood pressure (BP) through manual auscultation and the observation of Korotkoff sounds (KSs) remains the gold standard in BP methodology. Critical to determining BP levels via auscultation is the determination of KS audibility. While absolute sound level audibility is well researched, the problem has not been approached from the point of view of psychoacoustic masking of the sounds. Here, during manual auscultation of BP, a direct comparison is made between what an observer perceives as audible and the electronic analysis of audibility level determined from masking of sound signal levels. KSs are collected during auscultation with an electronic stethoscope, which allows simultaneously observing sound audibility and recording the sound electronically. By time-segmenting the recorded sound around Korotkoff peaks into a test segment and a masking segment, performing Fourier transforms on the segments, and comparing frequency-band sound energy levels, signal-to-noise ratios of a sound to its masking counterpart can be defined. Comparing these ratios to difference limen in the psychoacoustic masking literature, an approximate threshold for sound audibility is obtained. It is anticipated that this approach could have profound effects on future development of automated auscultation BP measurements.

Deep learning-based robust automatic non-invasive measurement of blood pressure using Korotkoff sounds

This paper proposes a method that automatically measures non-invasive blood pressure (BP) based on an auscultatory approach using Korotkoff sounds (K-sounds). There have been methods utilizing K-sounds that were more accurate in general than those using cuff pressure signals only under well-controlled environments, but most were vulnerable to the measurement conditions and to external noise because blood pressure is simply determined based on threshold values in the sound signal. The proposed method enables robust and precise BP measurements by evaluating the probability that each sound pulse is an audible K-sound based on a deep learning using a convolutional neural network (CNN). Instead of classifying sound pulses into two categories, audible K-sounds and others, the proposed CNN model outputs probability values. These values in a Korotkoff cycle are arranged in time order, and the blood pressure is determined. The proposed method was tested with a dataset acquired in practice that occasionally contains considerable noise, which can degrade the performance of the threshold-based methods. The results demonstrate that the proposed method outperforms a previously reported CNN-based classification method using K-sounds. With larger amounts of various types of data, the proposed method can potentially achieve more precise and robust results.

A Novel Automated Blood Pressure Estimation Algorithm Using Sequences of Korotkoff Sounds

The use of automated non-invasive blood pressure (NIBP) measurement devices is growing, as they can be used without expertise, and BP measurement can be performed by patients at home. Non-invasive cuff-based monitoring is the dominant method for BP measurement. While the oscillometric technique is most common, a few automated NIBP measurement methods have been developed based on the auscultatory technique. Amongst artificial intelligence (AI) techniques, deep learning has received increasing attention in different fields due to its strength in data classification, and feature extraction problems. This paper proposes a novel automated AI-based technique for NIBP estimation from auscultatory waveforms (AWs) based on converting the NIBP estimation problem to a sequence-to-sequence classification problem. To do this, a sequence of segments was first formed by segmenting the AWs, and their corresponding decomposed detail, and approximation parts obtained by wavelet packet decomposition method, and extracting features from each segment. Then, a label was assigned to each segment, i.e. (i) between systolic, and diastolic segments, and (ii) otherwise, and a bidirectional long short term memory recurrent neural network (BiLSTM-RNN) was devised to solve the resulting sequence-to-sequence classification problem. Adopting a 5-fold cross-validation scheme, and using a data base of 350 NIBP recordings gave an average mean absolute error of 1.7±3.7 mmHg for systolic BP (SBP), and 3.4 ±5.0 mmHg for diastolic BP (DBP) relative to reference values. Based on the results achieved, and comparisons made with the existing literature, it is concluded that the proposed automated BP estimation algorithm based on deep learning methods, and auscultatory waveform brings plausible benefits to the field of BP estimation.

Evaluation of cuff deflation and inflation rates on a deep learning-based automatic blood pressure measurement method: a pilot evaluation study

OBJECTIVE

The aim of this study was to evaluate the performance of using a deep learning-based method for measuring SBPs and DBPs and the effects of cuff inflation and deflation rates on the deep learning-based blood pressure (BP) measurement (in comparison with the manual auscultatory method).

METHODS

Forty healthy subjects were recruited. SBP and DBP were measured under four conditions (i.e. standard deflation, fast deflation, slow inflation and fast inflation) using both our newly developed deep learning-based method and the reference manual auscultatory method. The BPs measured under each condition were compared between the two methods. The performance of using the deep learning-based method to measure BP changes was also evaluated.

RESULTS

There were no significant BP differences between the two methods (P > 0.05), except for the DBPs measured during the slow and fast inflation conditions. By applying the deep learning-based method, SBPs measured from fast deflation, slow inflation and fast inflation decreased significantly by 3.0, 3.5 and 4.7 mmHg (all P < 0.05), respectively, in comparison with the standard deflation condition. Whereas, corresponding DBPs measured from the slow and fast inflation conditions increased significantly by 5.0 and 6.8 mmHg, respectively (both P < 0.05). There were no significant differences in BP changes measured by the two methods in most cases (all P > 0.05, except for DBP change in the slow and fast inflation conditions).

CONCLUSION

This study demonstrated that the deep learning-based method can achieve accurate BP measurement under the deflation and inflation conditions with different rates.

Subpleural Lung Changes Adjacent to Enlarged Internal Mammary Artery in Patients with Coarctation of Aorta and Pulmonary Hypoplasia: Some Thoughts on an Interesting Observation

Irregular pleural interface, minimally reduced lung density and/or focal subpleural cystic lung changes were noted in two patients with coarctation of aorta and in a patient with the right pulmonary hypoplasia. Lesions were distributed in the anterior segments of upper lobes, adjacent to internal mammary arteries (IMA). In view of unusual location of lung changes with unique distribution pattern prompted us to look for ‘cause and effect’ relation of events specific to long standing vascular pulsations, thus explaining the lung changes. While there are multiple factors for cystic lung disease, special distribution the lung changes seen in our patients leads us to explore mechanical process such as water hammer effect by dilated tortuous pulsatile IMA on the lungs as an additional etiology. A brief note on clinical and imaging appearance of our patients and discussion regarding water hammer theory are presented.

Quantitative Comparison of Korotkoff Sound Waveform Characteristics: Effects of Static Cuff Pressures and Stethoscope Positions

The underlying principles of Korotkoff sound (KorS) during blood pressure measurement and its waveform characteristic changes with cuff pressure and stethoscope position have not been fully understood. This study aimed to quantify the effects of cuff pressure and stethoscope position on the measured KorS waveform characteristics. Thirty healthy subjects were recruited in this study. Four stethoscopes were placed on the circumferential direction around the arm (m1, m2, m3 and m4; m1 was above the artery, and equal distance between each other), and then sequentially at three different longitudinal positions ('upper', 'middle' and 'low' part under the cuff). At each longitudinal position, three levels of static cuff pressure (high: SBP + 10 mmHg, low: DBP-10 mmHg, and medium: DBP + (SBP-DBP)/3) were applied during the recording of KorS waveform. The averaged KorS waveform was firstly computed by using an interpolation method, separately for measurements from different stethoscope locations and cuff pressures. Two quantitative indices were derived to characterize the recorded KorS waveform: intensity amplitude and high-level duration of KorS waveform. Post-hoc pairwise comparisons after analysis of variance were used to compare the waveform characteristic differences between different stethoscope locations and between cuff pressures. Variance analysis demonstrated that the effects of stethoscope circumferential and longitudinal positions and cuff pressure on the two KorS waveform indices were significant (all p < 0.001). In detail, KorS waveform recorded at cuff pressure P_MEDIUM had larger intensity amplitude and shorter high-level duration than those recorded at cuff pressure P_HIGH or P_LOW. In most conditions, the stethoscope above the artery (m1) produced the largest RMS intensity amplitude and shortest high-level duration, while the stethoscope at the opposite location of m1 generated the smallest RMS intensity amplitude and longest high-level duration. In terms of the effect of longitudinal position, the stethoscopes below the middle of the cuff always produced KorS recordings with larger intensity amplitude and shorter high-level duration. This study has quantified and provided scientific evidence that cuff pressure, stethoscope longitudinal and circumferential positions are important factors influencing KorS waveform characteristics.

Variation of the Korotkoff Stethoscope Sounds During Blood Pressure Measurement: Analysis Using a Convolutional Neural Network

Korotkoff sounds are known to change their characteristics during blood pressure (BP) measurement, resulting in some uncertainties for systolic and diastolic pressure (SBP and DBP) determinations. The aim of this study was to assess the variation of Korotkoff sounds during BP measurement by examining all stethoscope sounds associated with each heartbeat from above systole to below diastole during linear cuff deflation. Three repeat BP measurements were taken from 140 healthy subjects (age 21 to 73 years; 62 female and 78 male) by a trained observer, giving 420 measurements. During the BP measurements, the cuff pressure and stethoscope signals were simultaneously recorded digitally to a computer for subsequent analysis. Heartbeats were identified from the oscillometric cuff pressure pulses. The presence of each beat was used to create a time window (1 s, 2000 samples) centered on the oscillometric pulse peak for extracting beat-by-beat stethoscope sounds. A time-frequency two-dimensional matrix was obtained for the stethoscope sounds associated with each beat, and all beats between the manually determined SBPs and DBPs were labeled as "Korotkoff." A convolutional neural network was then used to analyze consistency in sound patterns that were associated with Korotkoff sounds. A 10-fold cross-validation strategy was applied to the stethoscope sounds from all 140 subjects, with the data from ten groups of 14 subjects being analyzed separately, allowing consistency to be evaluated between groups. Next, within-subject variation of the Korotkoff sounds analyzed from the three repeats was quantified, separately for each stethoscope sound beat. There was consistency between folds with no significant differences between groups of 14 subjects (P = 0.09 to P = 0.62). Our results showed that 80.7% beats at SBP and 69.5% at DBP were analyzed as Korotkoff sounds, with significant differences between adjacent beats at systole (13.1%, P = 0.001) and diastole (17.4%, P < 0.001). Results reached stability for SBP (97.8%, at sixth beat below SBP) and DBP (98.1%, at sixth beat above DBP) with no significant differences between adjacent beats (SBP P = 0.74; DBP P = 0.88). There were no significant differences at high-cuff pressures, but at low pressures close to diastole there was a small difference (3.3%, P = 0.02). In addition, greater within subject variability was observed at SBP (21.4%) and DBP (28.9%), with a significant difference between both (P < 0.02). In conclusion, this study has demonstrated that Korotkoff sounds can be consistently identified during the period below SBP and above DBP, but that at systole and diastole there can be substantial variations that are associated with high variation in the three repeat measurements in each subject.

A method for extracting respiratory frequency during blood pressure measurement, from oscillometric cuff pressure pulses and Korotkoff sounds recorded during the measurement

Respiratory frequency is an important physiological feature commonly used to assess health. However, the current measurements involve dedicated devices which not only increase the medical cost but also make health monitoring inconvenient. Earlier studies have shown that respiratory frequency could be extracted from electrocardiography (ECG) signal, but little was done to assess the possibility of extracting respiratory frequency from oscillometric cuff pressure pulses (OscP) or Korotkoff sounds (KorS), which are normally used for measuring blood pressure and more easily accessible than the ECG signal. This study presented a method to extract respiratory frequency from OscP and KorS during clinical blood pressure measurement. The method was evaluated with clinical data collected from 15 healthy participants, and its measurement accuracy was compared with a reference respiratory rate obtained with a magnetometer. Experimental results showed small non-significant mean absolute bias (0.019 Hz for OscP and 0.024 Hz for KorS) and high correlation (0.7 for both OscP and KorS) between the reference respiratory frequency and respiratory frequency extracted from OscP or KorS, indicating the high reliability of extracting respiratory frequency from OscP and KorS during normal blood pressure measurement.

Sources of inaccuracy in the measurement of adult patients' resting blood pressure in clinical settings: a systematic review

BACKGROUND

To interpret blood pressure (BP) data appropriately, healthcare providers need to be knowledgeable of the factors that can potentially impact the accuracy of BP measurement and contribute to variability between measurements.

METHODS

A systematic review of studies quantifying BP measurement inaccuracy. Medline and CINAHL databases were searched for empirical articles and systematic reviews published up to June 2015. Empirical articles were included if they reported a study that was relevant to the measurement of adult patients' resting BP at the upper arm in a clinical setting (e.g. ward or office); identified a specific source of inaccuracy; and quantified its effect. Reference lists and reviews were searched for additional articles.

RESULTS

A total of 328 empirical studies were included. They investigated 29 potential sources of inaccuracy, categorized as relating to the patient, device, procedure or observer. Significant directional effects were found for 27; however, for some, the effects were inconsistent in direction. Compared with true resting BP, significant effects of individual sources ranged from -23.6 to +33 mmHg SBP and -14 to +23 mmHg DBP.

CONCLUSION

A single BP value outside the expected range should be interpreted with caution and not taken as a definitive indicator of clinical deterioration. Where a measurement is abnormally high or low, further measurements should be taken and averaged. Wherever possible, BP values should be recorded graphically within ranges. This may reduce the impact of sources of inaccuracy and reduce the scope for misinterpretations based on small, likely erroneous or misleading, changes.

Assessment of arterial stiffness from pedal artery Korotkoff sound recordings: feasibility and potential utility

Brachial artery (BA) Korotkoff sound (KS) timing reflects arterial stiffness. We recorded pedal artery (PA) KS in 68 healthy subjects using an electronic stethoscope and electrocardiography. Intervals between QRS complex of the electrocardiogram and KS waveform peaks (termed the QKD interval) were measured for 60 seconds, averaged, and QKD velocity (v) calculated. Carotid-BA and carotid-PA pulse wave velocities (PWVs) were measured by applanation tonometry. Analyzable KS recordings were obtained from BA and PA in 100% and 92% subjects. PA QKDv decreased less than BA QKDv with progressive cuff inflation. At diastolic blood pressure + 20 mm Hg (maximal yield), BA QKDv was independently associated with weight and pulse pressure, whereas PA QKDv was related to weight and age. PA QKDv correlated with its corresponding PWV stronger than BA QKDv. In conclusion, PA KS is optimally recorded at diastolic blood pressure + 20 mm Hg; PA QKDv is correlated with age and better correlates with PWV than does BA QKDv. This technique may provide a simple arterial stiffness measure.

Effect of respiration on Korotkoff sounds and oscillometric cuff pressure pulses during blood pressure measurement

Blood pressure (BP) measurement accuracy depends on consistent changes in Korotkoff sounds (KorS) for manual measurement and oscillometric pulses for automated measurement, yet little is known about the direct effect of respiration on these physiological signals. The aim of this research was to quantitatively assess the modulation effect of respiration on Korotkoff sounds and oscillometric pulses. Systolic and diastolic blood pressures were measured manually from 30 healthy subjects (age 41 ± 12 years). Three static cuff pressure conditions were studied for two respiratory rates. Cuff pressure [with oscillometric pulses (OscP)], ECG, chest motion respiration [respiration signal (Resp), from magnetometer] and Korotkoff sounds (KorS, from digital stethoscope) were recorded twice for 20 s. The physiological data were evenly resampled. Respiratory frequency was calculated from Resp (f_R), OscP (f_O) and KorS (f_K) from peak spectral frequency. There was no statistically significant difference between f_R and f_O or f_K. Respiratory modulation was observed in all subjects. OscP amplitude modulation changed significantly between the two respiratory rates (p < 0.05) and between the three cuff pressures (p < 0.0001), and decreased significantly with decreasing cuff pressure (p < 0.05). The phase shift between Resp and modulation of OscP was statistically significant with respiratory rates (p < 0.05), but not with cuff pressures. It is accepted that BP in individuals is variable and that this relates to respiration; we now show that this respiration modulates oscillometric pulse and Korotkoff sound amplitudes from which BP is measured.

Improving auscultatory blood pressure measurement with electronic and computer technology: the visual auscultation method

BACKGROUND

The auscultatory method is used as the reference standard in all prevalent protocols for validation of noninvasive blood pressure measuring devices, and a validation study is essentially based on the comparison between the device and observer measurements. Thus, the objectivity and accuracy of observer measurements are crucial to the validation result.

METHODS

To provide observers with more objective information about an auscultatory measurement so that sufficient information to make measurements with greater potential objectivity and accuracy can be available, a computerized data acquisition and analysis system has been developed. It cannot only acquire and store Korotkoff sound, cuff pressure, and oscillometric pulse signals, as well as the sphygmomanometer image, but it also can display the waveforms of the three signals and the sphygmomanometer video while playing the synchronous Korotkoff sounds. With this system, observers can make their measurements via the visual auscultation method, that is to say, by watching those waveforms, instead of the sphygmomanometer, while listening for synchronized Korotkoff sounds. The system was validated according to the International Protocol (IP).

RESULTS

The result showed that all the differences between system measurements by the visual auscultation method and observer measurements by the conventional auscultatory method were within 4 mm Hg.

CONCLUSIONS

The visual auscultation method achieved a high degree of accuracy, and human observers can be replaced by the system in the validation study of blood pressure measuring devices.