Development of a graphical user interface for automatic separation of human voice from Doppler ultrasound audio in diving experiments.

Doppler ultrasound (DU) is used in decompression research to detect venous gas emboli in the precordium or subclavian vein, as a marker of decompression stress. This is of relevance to scuba divers, compressed air workers and astronauts to prevent decompression sickness (DCS) that can be caused by these bubbles upon or after a sudden reduction in ambient pressure. Doppler ultrasound data is graded by expert raters on the Kisman-Masurel or Spencer scales that are associated to DCS risk. Meta-analyses, as well as efforts to computer-automate DU grading, both necessitate access to large databases of well-curated and graded data. Leveraging previously collected data is especially important due to the difficulty of repeating large-scale extreme military pressure exposures that were conducted in the 70-90s in austere environments. Historically, DU data (Non-speech) were often captured on cassettes in one-channel audio with superimposed human speech describing the experiment (Speech). Digitizing and separating these audio files is currently a lengthy, manual task. In this paper, we develop a graphical user interface (GUI) to perform automatic speech recognition and aid in Non-speech and Speech separation. This constitutes the first study incorporating speech processing technology in the field of diving research. If successful, it has the potential to significantly accelerate the reuse of previously-acquired datasets. The recognition task incorporates the Google speech recognizer to detect the presence of human voice activity together with corresponding timestamps. The detected human speech is then separated from the audio Doppler ultrasound within the developed GUI. Several experiments were conducted on recently digitized audio Doppler recordings to corroborate the effectiveness of the developed GUI in recognition and separations tasks, and these are compared to manual labels for Speech timestamps. The following metrics are used to evaluate performance: the average absolute differences between the reference and detected Speech starting points, as well as the percentage of detected Speech over the total duration of the reference Speech. Results have shown the efficacy of the developed GUI in Speech/Non-speech component separation.

An open-source framework for synthetic post-dive Doppler ultrasound audio generation.

Doppler ultrasound (DU) measurements are used to detect and evaluate venous gas emboli (VGE) formed after decompression. Automated methodologies for assessing VGE presence using signal processing have been developed on varying real-world datasets of limited size and without ground truth values preventing objective evaluation. We develop and report a method to generate synthetic post-dive data using DU signals collected in both precordium and subclavian vein with varying degrees of bubbling matching field-standard grading metrics. This method is adaptable, modifiable, and reproducible, allowing for researchers to tune the produced dataset for their desired purpose. We provide the baseline Doppler recordings and code required to generate synthetic data for researchers to reproduce our work and improve upon it. We also provide a set of pre-made synthetic post-dive DU data spanning six scenarios representing the Spencer and Kisman-Masurel (KM) grading scales as well as precordial and subclavian DU recordings. By providing a method for synthetic post-dive DU data generation, we aim to improve and accelerate the development of signal processing techniques for VGE analysis in Doppler ultrasound.

The risk of decompression illness in breath-hold divers: a systematic review.

INTRODUCTION: Breath-hold (BH) diving has known risks, for example drowning, pulmonary oedema of immersion and barotrauma. There is also the risk of decompression illness (DCI) from decompression sickness (DCS) and/or arterial gas embolism (AGE). The first report on DCS in repetitive freediving was published in 1958 and from then there have been multiple case reports and a few studies but no prior systematic review or meta-analysis. METHODS: We undertook a systematic literature review to identify articles available from PubMed and Google Scholar concerning breath-hold diving and DCI up to August 2021. RESULTS: The present study identified 17 articles (14 case reports, three experimental studies) covering 44 incidences of DCI following BH diving. CONCLUSIONS: This review found that the literature supports both DCS and AGE as potential mechanisms for DCI in BH divers; both should be considered a risk for this cohort of divers, just as for those breathing compressed gas while underwater.

A fully automated algorithm for heart rate detection in post-dive precordial Doppler ultrasound.

Background: Doppler ultrasound is used currently in decompression research for the evaluation of venous gas emboli (VGE). Estimation of heart rate from post-dive Doppler ultrasound recordings can provide a tool for the evaluation of physiological changes from decompression stress, as well as aid in the development of automated VGE detection algorithms that relate VGE presence to cardiac activity. Method: An algorithm based on short-term autocorrelation was developed in MATLAB to estimate the heart rate in post-dive precordial Doppler ultrasound. The algorithm was evaluated on 21 previously acquired and labeled precordial recordings spanning Kisman-Masurel (KM) codes of 111-444 (KM I-IV) with manually derived instantaneous heart rates. Results: A window size of at least two seconds was necessary for robust and accurate instantaneous heart rate estimation with a mean error of 1.56 ± 7.10 bpm. Larger window sizes improved the algorithm performance, at the cost of beat-to-beat heart rate estimates. We also found that our algorithm provides good results for low KM grade Doppler recordings with and without flexion, and high KM grades without flexion. High KM grades observed after movement produced the greatest mean absolute error of 6.12 ± 8.40 bpm. Conclusion: We have developed a fully automated algorithm for the estimation of heart rate in post-dive precordial Doppler ultrasound recordings.

Deep Learning-Based Venous Gas Emboli Grade Classification in Doppler Ultrasound Audio Recordings.

OBJECTIVE: Doppler ultrasound (DU) is used to detect venous gas emboli (VGE) post dive as a marker of decompression stress for diving physiology research as well as new decompression procedure validation to minimize decompression sickness risk. In this article, we propose the first deep learning model for VGE grading in DU audio recordings. METHODS: A database of real-world data was assembled and labeled for the purpose of developing the algorithm, totaling 274 recordings comprising both subclavian and precordial measurements. Synthetic data was also generated by acquiring baseline DU signals from human volunteers and superimposing laboratory-acquired DU signals of bubbles flowing in a tissue mimicking material. A novel squeeze-and-excitation deep learning model was designed to effectively classify recordings on the 5-class Spencer scoring system used by trained human raters. RESULTS: On the real-data test set, we show that synthetic data pretraining achieves average ordinal accuracy of 84.9% for precordial and 90.4% for subclavian DU which is a 24.6% and 26.2% increase over training with real-data and time-series augmentation only. The weighted kappa coefficients of agreement between the model and human ground truth were 0.74 and 0.69 for precordial and subclavian respectively, indicating substantial agreement similar to human inter-rater agreement for this type of data. CONCLUSION: The present work demonstrates the first application of deep-learning for DU VGE grading using a combination of synthetic and real-world data. SIGNIFICANCE: The proposed method can contribute to accelerating DU analysis for decompression research.

Observed decompression sickness and venous bubbles following 18-msw dive profiles using RN Table 11.

The venous bubble load in the body after diving may be used to infer risk of decompression sickness (DCS). Retrospective analysis of post-dive bubbling and DCS was made on seven studies. Each of these investigated interventions, using an 18 meters of sea water (msw) air dive profile from Royal Navy Table 11 (Mod Air Table), equivalent to the Norwegian Air tables. A recent neurological DCS case suggested this table was not safe as thought. Two-hundred and twenty (220) man-dives were completed on this profile. Bubble measurements were made following 219 man-dives, using Doppler or 2D ultrasound measurements made on the Kisman-Masurel and Eftedal-Brubakk scales, respectively. The overall median grade was KM/EB 0.5 and the overall median maximum grade was KM/EB 2. Two cases of transient shoulder discomfort ("niggles") were observed (0.9% (95% CL 0.1% - 3.3%)) and were treated with surface oxygen. One dive, for which no bubble measurements were made, resulted in a neurological DCS treated with hyperbaric oxygen. The DCS risk of this profile is below that predicted by models, and comparison of the cumulative incidence of DCS of these data to the large dataset compiled by DCIEM [1, 2] show that the incidence is lower than might be expected.

Consensus guidelines for the use of ultrasound for diving research.

The International Meeting on Ultrasound for Diving Research produced expert consensus recommendations for ultrasound detection of vascular gas bubbles and the analysis, interpretation and reporting of such data. Recommendations for standardization of techniques to allow comparison between studies included bubble monitoring site selection, frequency and duration of monitoring, and use of the Spencer, Kisman-Masurel or Eftedal-Brubakk scales. Recommendations for reporting of results included description of subject posture and provocation manoeuvres during monitoring, reporting of untransformed data and the appropriate use of statistics. These guidelines are available from www.dhmjournal.com.

Ultrasound detection of vascular decompression bubbles: the influence of new technology and considerations on bubble load.

INTRODUCTION: Diving often causes the formation of 'silent' bubbles upon decompression. If the bubble load is high, then the risk of decompression sickness (DCS) and the number of bubbles that could cross to the arterial circulation via a pulmonary shunt or patent foramen ovale increase. Bubbles can be monitored aurally, with Doppler ultrasound, or visually, with two dimensional (2D) ultrasound imaging. Doppler grades and imaging grades can be compared with good agreement. Early 2D imaging units did not provide such comprehensive observations as Doppler, but advances in technology have allowed development of improved, portable, relatively inexpensive units. Most now employ harmonic technology; it was suggested that this could allow previously undetectable bubbles to be observed. METHODS: This paper provides a review of current methods of bubble measurement and how new technology may be changing our perceptions of the potential relationship of these measurements to decompression illness. Secondly, 69 paired ultrasound images were made using conventional 2D ultrasound imaging and harmonic imaging. Images were graded on the Eftedal-Brubakk (EB) scale and the percentage agreement of the images calculated. The distribution of mismatched grades was analysed. RESULTS: Fifty-four of the 69 paired images had matching grades. There was no significant difference in the distribution of high or low EB grades for the mismatched pairs. CONCLUSIONS: Given the good level of agreement between pairs observed, it seems unlikely that harmonic technology is responsible for any perceived increase in observed bubble loads, but it is probable that our increasing use of 2D ultrasound to assess dive profiles is changing our perception of 'normal' venous and arterial bubble loads. Methods to accurately investigate the load and size of bubbles developed will be helpful in the future in determining DCS risk.

Oxygen and carbogen breathing following simulated submarine escape.

Escape from a disabled submarine exposes escapers to a high risk of decompression sickness (DCS). The initial bubble load is thought to emanate from the fast tissues; it is this load that should be lowered to reduce risk of serious neurological DCS. The breathing of oxygen or carbogen (5% CO2, 95% O2) post-surfacing was investigated with regard to its ability to reduce the initial bubble load in comparison to air breathing. Thirty-two goats were subject to a dry simulated submarine escape profile to and from 240 meters (2.5 MPa). On surfacing, they breathed air (control), oxygen or carbogen for 30 minutes. Regular Doppler audio bubble grading was carried out, using the Kisman Masurel (KM) scale. One suspected case of DCS was noted. No oxygen toxicity or arterial gas embolism occurred. No significant difference was found between the groups in terms of the median peak KM grade or the period before the KM grade dropped below III. Time to disappearance of bubbles was significantly different between groups; oxygen showed faster bubble resolution than carbogen and air. This reduction in time to bubble resolution may be beneficial in reducing decompression stress, but probably does not affect the risk of fast-tissue DCS.

Pre-dive exercise and post-dive evolution of venous gas emboli.

BACKGROUND: Recent studies have indicated that exercise before diving significantly reduces the number of circulating bubbles and the risk of decompression sickness. However, the most effective time delay between exercise and dive is not clear; the present aim was to resolve this. METHODS: In a hyperbaric chamber, 10 men were compressed to 18 m for 100 min, then decompressed as per Royal Navy Table 11. Each subject performed three dives: a control dive and two after exercise performed either 24 h or 2 h before diving. Exercise consisted of 40 min submaximal work on a cycle ergometer. Venous gas emboli (VGE) were evaluated using precordial Doppler ultrasound immediately on surfacing, with measurements made at 5-min intervals for 30 min, and at 15-min intervals for at least 2.5 h total using the Kisman Masurel (KM) scale. RESULTS: Exercise either 24 or 2 h prior to a dive did not reduce the median number of circulating VGE (median maximum KM grade: control, 2+; for both exercise dives, 3). Bubbles disappeared from the circulation faster after the control dive than the exercise dives. Time to median KM Doppler scores of zero were: control:120 min; 2-h group: 225 min; 24-h group: 165 min. CONCLUSION: Cycling exercise prior to diving did not reduce the number of circulating VGE in comparison to control, in contrast to recent studies. A number of factors may be responsible for these findings, including type of exercise performed, wet diving experience, and disparity in Doppler measurement techniques.

The need for optimisation of post-dive ultrasound monitoring to properly evaluate the evolution of venous gas emboli.

Audio Doppler ultrasound and echocardiographic techniques are useful tools for investigating the formation of inert gas bubbles after hyperbaric exposure and can help to assess the risk of occurrence of decompression sickness. However, techniques, measurement period and regularity of measurements must be standardised for results to be comparable across research groups and to be of any benefit. There now appears to be a trend for fewer measurements to be made than recommended, which means that the onset, peak and cessation of bubbling may be overlooked and misreported. This review summarises comprehensive Doppler data collected over 15 years across many dive profiles and then assesses the effectiveness of measurements made between 30 and 60 minutes (min) post-dive (commonly measured time points made in recent studies) in characterising the evolution and peak of venous gas emboli (VGE). VGE evolution in this dive series varied enormously both intra- and inter-individually and across dive profiles. Median, rather than mean values are best reported when describing data which have a non-linear relation to the underlying number of bubbles, as are median peak grades, rather than maximum, which may reflect only one individual's data. With regard to monitoring, it is apparent that the evolution of VGE cannot be described across multiple dive profiles using measurements made at only 30 to 60 min, or even 90 min post-dive. Earlier and more prolonged measurement is recommended, while the frequency of measurements should also be increased; in doing so, the accuracy and value of studies dependent on bubble evolution will be improved.

Venous gas emboli and exhaled nitric oxide with simulated and actual extravehicular activity.

The decompression experienced due to the change in pressure from a space vehicle (1013hPa) to that in a suit for extravehicular activity (EVA) (386hPa) was simulated using a hypobaric chamber. Previous ground-based research has indicated around a 50% occurrence of both venous gas emboli (VGE) and symptoms of decompression illness (DCI) after similar decompressions. In contrast, no DCI symptoms have been reported from past or current space activities. Twenty subjects were studied using Doppler ultrasound to detect any VGE during decompression to 386hPa, where they remained for up to 6h. Subjects were supine to simulate weightlessness. A large number of VGE were found in one subject at rest, who had a recent arm fracture; a small number of VGE were found in another subject during provocation with calf contractions. No changes in exhaled nitric oxide were found that can be related to either simulated EVA or actual EVA (studied in a parallel study on four cosmonauts). We conclude that weightlessness appears to be protective against DCI and that exhaled NO is not likely to be useful to monitor VGE.

Pulse oximetry to detect hypoxemia during apnea: comparison of finger and ear probes.

INTRODUCTION: When investigating apnea, for example in diving or altitude studies, hypoxemia is a variable that must be monitored to reduce the risk of hypoxic syncope. Pulse oximetry is a simple technique that measures arterial oxygen saturation (SpO2). As apnea induces a peripheral vasoconstriction, we hypothesized that it would be better to measure hypoxia using more centrally placed ear lobe oximetry probes rather than peripheral finger probes. METHODS: Seven men were studied, ages 18-35. Two pulse oximeters were used, a Satlite Trans (Ox-1) and Ohmeda Biox (Ox-2), both with ear and finger probes. Subjects carried out a sub-maximal breath hold for 60 s while performing dynamic leg exercise on a cycle ergometer at 50 W. Subjects performed the maneuver six times in total, in a crossover design. RESULTS: The Ox-1 finger probe showed 6.0 +/- 3.7% higher values than the ear-lobe probe at their respective nadirs. The Ox-2 probes differed in the same manner by 6.5 +/- 4.2%. The average delay between the nadir shown by the ear and finger probes was 15 s (+/- 3.5). When the ear-probes were at their nadir (SpO2 78 +/- 3.5%), the finger probes had considerably higher SpO2 levels (94.6 +/- 3.5%). DISCUSSION: Apneic induced hypoxemia was monitored poorly by finger probe pulse oximetry. The delay in response may jeopardize safety, for example in breath-hold diving studies. Hypoxemia does not seem to be accurately reflected by finger measurements in situations where peripheral vasoconstriction may occur.

Cerebral blood flow velocity and psychomotor performance during acute hypoxia.

INTRODUCTION: The physiological effects of hypoxic environments can help determine safe limits for workers where cognitive and motor performance is important. We investigated the effects of a PIO2 of 15 kPa and 10 kPa on medial cerebral artery blood flow velocity (CBFV) and psychomotor performance. METHODS: Over 3 sessions, each involving 3 separate test batteries, 13 subjects breathed either 21 kPa PIO2 (control), 15 kPa PIO2, or 10 kPa PIO2. The tests measured reaction time, spatial orientation, voluntary repetitive movement, and fine manipulation. CBFV, PETCO2, PETO2, Sa02, and BP were recorded throughout. RESULTS: ANOVA analysis showed that 15 kPa PIO2 did not significantly change psychomotor test performance. The mean number of incorrect responses in the reaction time test significantly increased to 5.6 (SD - 4.0) while breathing 10 kPa PIO2, as did the mean number of errors (7.7 +/- 5.0) in the fine manipulation test. Only 10 kPa PIO2 affected CBFV, causing a significant increase in flow from 50 +/- 6.5 cm x s(-1) to 55 +/- 10.3 cm x s(-1). CBFV significantly increased during three psychomotor tests while breathing air; however, it did not increase further during psychomotor testing in hypoxia. DISCUSSION: A PIo2 of 15 kPa did not affect subject performance, and should not cause operational risk. At 10 kPa PIO2, accuracy and vigilance were slightly affected; however, the reduction in oxygenation was not great enough to cause major decrements. CBFV was not a good indicator of mental stress during hypoxia.