ABSTRACT
BACKGROUND AND OBJECTIVE: Respiratory diseases are among the most significant causes of morbidity and mortality worldwide, causing substantial strain on society and health systems. Over the last few decades, there has been increasing interest in the automatic analysis of respiratory sounds and electrical impedance tomography (EIT). Nevertheless, no publicly available databases with both respiratory sound and EIT data are available. METHODS: In this work, we have assembled the first open-access bimodal database focusing on the differential diagnosis of respiratory diseases (BRACETS: Bimodal Repository of Auscultation Coupled with Electrical Impedance Thoracic Signals). It includes simultaneous recordings of single and multi-channel respiratory sounds and EIT. Furthermore, we have proposed several machine learning-based baseline systems for automatically classifying respiratory diseases in six distinct evaluation tasks using respiratory sound and EIT (A1, A2, A3, B1, B2, B3). These tasks included classifying respiratory diseases at sample and subject levels. The performance of the classification models was evaluated using a 5-fold cross-validation scheme (with subject isolation between folds). RESULTS: The resulting database consists of 1097 respiratory sounds and 795 EIT recordings acquired from 78 adult subjects in two countries (Portugal and Greece). In the task of automatically classifying respiratory diseases, the baseline classification models have achieved the following average balanced accuracy: Task A1 - 77.9±13.1%; Task A2 - 51.6±9.7%; Task A3 - 38.6±13.1%; Task B1 - 90.0±22.4%; Task B2 - 61.4±11.8%; Task B3 - 50.8±10.6%. CONCLUSION: The creation of this database and its public release will aid the research community in developing automated methodologies to assess and monitor respiratory function, and it might serve as a benchmark in the field of digital medicine for managing respiratory diseases. Moreover, it could pave the way for creating multi-modal robust approaches for that same purpose.
Subject(s)
Respiration , Respiratory Tract Diseases , Thorax , Auscultation/instrumentation , Thorax/physiology , Electric Impedance , Humans , Male , Middle Aged , Aged , Adult , Respiratory Tract Diseases/diagnosis , Respiratory Tract Diseases/physiopathologyABSTRACT
BACKGROUND: Very low birth weight (VLBW) infants on noninvasive ventilation (NIV) experience frequent fluctuations in oxygen saturation (SpO2) that are associated with an increased risk for mortality and severe morbidities. METHODS: In this randomized crossover trial, VLBW infants (n = 22) born 22+3 to 28+0 weeks on NIV with supplemental oxygen were allocated on two consecutive days in random order to synchronized nasal intermittent positive pressure ventilation (sNIPPV) and nasal high-frequency oscillatory ventilation (nHFOV) for 8 h. nHFOV and sNIPPV were set to equivalent mean airway pressure and transcutaneous pCO2. Primary outcome was the time spent within the SpO2 target (88-95%). RESULTS: During sNIPPV, VLBW infants spent significantly more time within the SpO2 target (59.9%) than during nHFOV (54.6%). The proportion of time spent in hypoxemia (22.3% vs. 27.1%) and the mean fraction of supplemental oxygen (FiO2) (29.4% vs. 32.8%) were significantly reduced during sNIPPV, while the respiratory rate (50.1 vs. 42.6) was significantly higher. Mean SpO2, SpO2 above the target, number of prolonged (>1 min) and severe (SpO2 <80%) hypoxemic episodes, parameters of cerebral tissue oxygenation using NIRS, number of FiO2 adjustments, heart rate, number of bradycardias, abdominal distension and transcutaneous pCO2 did not differ between both interventions. CONCLUSIONS: In VLBW infants with frequent fluctuations in SpO2, sNIPPV is more efficient than nHFOV to retain the SpO2 target and to reduce FiO2 exposure. These results demand more detailed investigations into cumulative oxygen toxicities during different modes of NIV over the weaning period, particularly with regard to consequences for long-term outcomes.
Subject(s)
High-Frequency Ventilation , Noninvasive Ventilation , Infant, Newborn , Infant , Humans , Intermittent Positive-Pressure Ventilation/methods , Infant, Premature , Oxygen Saturation , Cross-Over Studies , Infant, Very Low Birth Weight , Noninvasive Ventilation/methods , Oxygen , Continuous Positive Airway Pressure/methodsABSTRACT
Objective.Viral lower respiratory tract infections (LRTI) are the leading cause for acute admission to the intensive care unit in infants and young children. Nebulized bronchodilators are often used when treating the most severe cases. The aim of this study was to investigate the bronchodilator effect on respiratory mechanics during intensive care with electrical impedance tomography (EIT) and to assess the feasibility of EIT in this context.Approach.We continuously monitored the children with chest EIT for up to 72 h in an observational study design. The treatment decisions were done by clinical assessment, as the clinicians were blinded to the EIT information during data collection. In a retrospective analysis, clinical parameters and regional expiratory time constants determined by EIT were used to assess the effects of bronchodilator administration, especially regarding airway resistance.Main results.We included six children from 11 to 27 months of age requiring intensive care due to viral LRTI and receiving bronchodilator agents. Altogether 131 bronchodilator administrations were identified during EIT monitoring. After validation of the exact timing of events and EIT data quality, 77 administrations were included in the final analysis. Fifty-five bronchodilator events occurred during invasive ventilation and 22 during high-flow nasal cannulae treatment. Only 17% of the bronchodilator administrations resulted in a relevant decrease in calculated expiratory time constants.Significance.Continuous monitoring with EIT might help to optimize the treatment of LRTI in pediatric intensive care units. In particular, EIT-based regional expiratory time constants would allow objective assessment of the effects of bronchodilators and other respiratory therapies.
Subject(s)
Bronchodilator Agents , Respiratory Tract Infections , Infant , Humans , Child , Child, Preschool , Bronchodilator Agents/pharmacology , Bronchodilator Agents/therapeutic use , Tomography/methods , Electric Impedance , Retrospective Studies , Lung , Respiratory Tract Infections/drug therapyABSTRACT
Rationale: Electrical impedance tomography (EIT) allows instantaneous and continuous visualization of regional ventilation and changes in end-expiratory lung volume at the bedside. There is particular interest in using EIT for monitoring in critically ill neonates and young children with respiratory failure. Previous studies have focused only on short-term monitoring in small populations. The feasibility and safety of prolonged monitoring with EIT in neonates and young children have not been demonstrated yet. Objectives: To evaluate the feasibility and safety of long-term EIT monitoring in a routine clinical setting and to describe changes in ventilation distribution and homogeneity over time and with positioning in a multicenter cohort of neonates and young children with respiratory failure. Methods: At four European University hospitals, we conducted an observational study (NCT02962505) on 200 patients with postmenstrual ages (PMA) between 25 weeks and 36 months, at risk for or suffering from respiratory failure. Continuous EIT data were obtained using a novel textile 32-electrode interface and recorded at 48 images/s for up to 72 hours. Clinicians were blinded to EIT images during the recording. EIT parameters and the effects of body position on ventilation distribution were analyzed offline. Results: The average duration of EIT measurements was 53 ± 20 hours. Skin contact impedance was sufficient to allow image reconstruction for valid ventilation analysis during a median of 92% (interquartile range, 77-98%) of examination time. EIT examinations were well tolerated, with minor skin irritations (temporary redness or imprint) occurring in 10% of patients and no moderate or severe adverse events. Higher ventilation amplitude was found in the dorsal and right lung areas when compared with the ventral and left regions, respectively. Prone positioning resulted in an increase in the ventilation-related EIT signal in the dorsal hemithorax, indicating increased ventilation of the dorsal lung areas. Lateral positioning led to a redistribution of ventilation toward the dependent lung in preterm infants and to the nondependent lung in patients with PMA > 37 weeks. Conclusions: EIT allows continuous long-term monitoring of regional lung function in neonates and young children for up to 72 hours with minimal adverse effects. Our study confirmed the presence of posture-dependent changes in ventilation distribution and their dependency on PMA in a large patient cohort. Clinical trial registered with www.clinicaltrials.gov (NCT02962505).
Subject(s)
Respiratory Insufficiency , Tomography , Child , Child, Preschool , Electric Impedance , Humans , Infant , Infant, Newborn , Infant, Premature , Lung/diagnostic imaging , Respiratory Insufficiency/diagnosis , Respiratory Insufficiency/therapy , Tomography/methodsABSTRACT
The aim of this study was to examine whether electrical impedance tomography (EIT) could determine the presence of ventilation inhomogeneity in patients with chronic obstructive lung disease (COPD) from measurements carried out not only during conventional forced full expiration maneuvers but also from forced inspiration maneuvers and quiet tidal breathing and whether the inhomogeneity levels were comparable among the phases and higher than in healthy subjects. EIT data were acquired in 52 patients with exacerbated COPD (11 women, 41 men, 68 ± 11 years) and 14 healthy subjects (6 women, 8 men, 38 ± 8 years). Regional lung function parameters of forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1), forced inspiratory vital capacity (FIVC), forced inspiratory volume in 1 s (FIV1), and tidal volume (V T ) were determined in 912 image pixels. The spatial inhomogeneity of the pixel parameters was characterized by the coefficients of variation (CV) and the global inhomogeneity (GI) index. CV and GI values of pixel FVC, FEV1, FIVC, FIV1, and VT were significantly higher in patients than in healthy subjects (p ≤ 0.0001). The ventilation distribution was affected by the analyzed lung function parameter in patients (CV: p = 0.0024, GI: p = 0.006) but not in healthy subjects. Receiver operating characteristic curves showed that CV and GI discriminated patients from healthy subjects with an area under the curve (AUC) of 0.835 and 0.852 (FVC), 0.845 and 0.867 (FEV1), 0.903 and 0.903 (FIVC), 0.891 and 0.882 (FIV1), and 0.821 and 0.843 (VT), respectively. These findings confirm the ability of EIT to identify increased ventilation inhomogeneity in patients with COPD.
ABSTRACT
Patients suffering from pulmonary diseases typically exhibit pathological lung ventilation in terms of homogeneity. Electrical Impedance Tomography (EIT) is a non- invasive imaging method that allows to analyze and quantify the distribution of ventilation in the lungs. In this article, we present a new approach to promote the use of EIT data and the implementation of new clinical applications for differential diagnosis, with the development of several machine learning models to discriminate between EIT data from healthy and nonhealthy subjects. EIT data from 16 subjects were acquired: 5 healthy and 11 non-healthy subjects (with multiple pulmonary conditions). Preliminary results have shown accuracy percentages of 66% in challenging evaluation scenarios. The results suggest that the pairing of EIT feature engineering methods with machine learning methods could be further explored and applied in the diagnostic and monitoring of patients suffering from lung diseases. Also, we introduce the use of a new feature in the context of EIT data analysis (Impedance Curve Correlation).
Subject(s)
Pulmonary Ventilation , Tomography , Electric Impedance , Humans , Machine Learning , Tomography, X-Ray ComputedABSTRACT
Electrical impedance tomography (EIT) is a noninvasive imaging modality that allows a continuous assessment of changes in regional bioimpedance of different organs. One of its most common biomedical applications is monitoring regional ventilation distribution in critically ill patients treated in intensive care units. In this work, we put forward a proof-of-principle study that demonstrates how one can reconstruct synchronously measured respiratory or circulatory parameters from the EIT image sequence using a deep learning model trained in an end-to-end fashion. For this purpose, we devise an architecture with a convolutional feature extractor whose output is processed by a recurrent neural network. We demonstrate that one can accurately infer absolute volume, absolute flow, normalized airway pressure and within certain limitations even the normalized arterial blood pressure from the EIT signal alone, in a way that generalizes to unseen patients without prior calibration. As an outlook with direct clinical relevance, we furthermore demonstrate the feasibility of reconstructing the absolute transpulmonary pressure from a combination of EIT and absolute airway pressure, as a way to potentially replace the invasive measurement of esophageal pressure. With these results, we hope to stimulate further studies building on the framework put forward in this work.
Subject(s)
Tomography, X-Ray Computed , Tomography , Electric Impedance , Humans , Intensive Care Units , Lung/diagnostic imagingABSTRACT
OBJECTIVE: We aim to provide an algorithm for the detection of myocardial infarction that operates directly on ECG data without any preprocessing and to investigate its decision criteria. APPROACH: We train an ensemble of fully convolutional neural networks on the PTB ECG dataset and apply state-of-the-art attribution methods. MAIN RESULTS: Our classifier reaches 93.3% sensitivity and 89.7% specificity evaluated using 10-fold cross-validation with sampling based on patients. The presented method outperforms state-of-the-art approaches and reaches the performance level of human cardiologists for detection of myocardial infarction. We are able to discriminate channel-specific regions that contribute most significantly to the neural network's decision. Interestingly, the network's decision is influenced by signs also recognized by human cardiologists as indicative of myocardial infarction. SIGNIFICANCE: Our results demonstrate the high prospects of algorithmic ECG analysis for future clinical applications considering both its quantitative performance as well as the possibility of assessing decision criteria on a per-example basis, which enhances the comprehensibility of the approach.