Intracerebral hemorrhage detection on computed tomography images using a residual neural network.

Intracerebral hemorrhage (ICH) is a high mortality rate, critical medical injury, produced by the rupture of a blood vessel of the vascular system inside the skull. ICH can lead to paralysis and even death. Therefore, it is considered a clinically dangerous disease that needs to be treated quickly. Thanks to the advancement in machine learning and the computing power of today's microprocessors, deep learning has become an unbelievably valuable tool for detecting diseases, in particular from medical images. In this work, we are interested in differentiating computer tomography (CT) images of healthy brains and ICH using a ResNet-18, a deep residual convolutional neural network. In addition, the gradient-weighted class activation mapping (Grad-CAM) technique was employed to visually explore and understand the network's decisions. The generalizability of the detector was assessed through a 100-iteration Monte Carlo cross-validation (80% of the data for training and 20% for test). In a database with 200 CT images of brains (100 with ICH and 100 without ICH), the detector yielded, on average, 95.93%accuracy, 96.20% specificity, 95.65% sensitivity, 96.40% precision, and 95.91% F1-core, with an average computing time of 165.90 s to train the network (on 160 images) and 1.17 s to test it with 40 CT images. These results are comparable with the state of the art with a simpler and lower computational load approach. Our detector could assist physicians in their medical decision, in resource optimization and in reducing the time and error in the diagnosis of ICH.

Hidden Markov model-based heartbeat detector using electrocardiogram and arterial pressure signals.

The automatic detection of a heartbeat is commonly performed by detecting the QRS complex in the electrocardiogram (ECG), however, various noise sources and missing data can jeopardize the reliability of the ECG. Therefore, there is a growing interest in combining the information from many physiological signals to accurately detect heartbeats. To this end, hidden Markov models (HMMs) are used in this work to jointly exploit the information from ECG, arterial blood pressure (ABP) and pulmonary arterial pressure (PAP) signals in order to conceive a heartbeat detector. After preprocessing the physiological signals, a sliding window is used to extract an observation sequence to be passed through two HMMs (previously trained on a training dataset) in order to obtain the log-likelihoods of observation and signals a detection if the difference of log-likelihoods exceeds an adaptive threshold. Several HMM-based heartbeat detectors were conceived to exploit the information from the ECG, ABP and PAP signals from the MIT-BIH Arrhythmia, PhysioNet Computing in Cardiology Challenge 2014, and MGH/MF Waveform databases. A grid search methodology was used to optimize the duration of the observation sequence and a multiplicative factor to form the adaptive threshold. Using the optimal parameters found on a training database through 10-fold cross-validation, sensitivity and positive predictivity above 99% were obtained on the MIT-BIH Arrhythmia and PhysioNet Computing in Cardiology Challenge 2014 databases, while they are above 95% in the MGH/MF waveform database using ECG and ABP signals. Our detector approach showed detection performances comparable with the literature in the three databases. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s13534-021-00192-x.

Early bradycardia detection and therapeutic interventions in preterm infant monitoring.

In very preterm infants, cardio-respiratory events and associated hypoxemia occurring during early postnatal life have been associated with risks of retinopathy, growth alteration and neurodevelopment impairment. These events are commonly detected by continuous cardio-respiratory monitoring in neonatal intensive care units (NICU), through the associated bradycardia. NICU nurse interventions are mainly triggered by these alarms. In this work, we acquired data from 52 preterm infants during NICU monitoring, in order to propose an early bradycardia detector which is based on a decentralized fusion of three detectors. The main objective is to improve automatic detection under real-life conditions without altering performance with respect to that of a monitor commonly used in NICU. We used heart rate lower than 80 bpm during at least 10 sec to define bradycardia. With this definition we observed a high rate of false alarms (64%) in real-life and that 29% of the relevant alarms were not followed by manual interventions. Concerning the proposed detection method, when compared to current monitors, it provided a significant decrease of the detection delay of 2.9 seconds, without alteration of the sensitivity (97.6% vs 95.2%) and false alarm rate (63.7% vs 64.1%). We expect that such an early detection will improve the response of the newborn to the intervention and allow for the development of new automatic therapeutic strategies which could complement manual intervention and decrease the sepsis risk.

Optimized fasting and OGTT-based simple surrogate methods for assessing insulin sensitivity.

AIMS: Simple surrogate indices of insulin sensitivity have been conceived to deal with costly and complicated approaches, such as the hyperinsulinemic-euglycemic clamp; however, their use has not been widespread given their variabilities in different populations. In this paper, we present two simple surrogate indices, one that uses fasting glucose and insulin values and the other based on the values from the oral glucose tolerance test. MATERIALS AND METHODS: The proposed methods integrate easy-to-obtain anthropometric measures. Evolutionary algorithms were used to optimize the proposed methods by maximizing its correlation with the Stumvoll MCR method. RESULTS AND CONCLUSION: When the proposed indices were applied to three study groups (control subjects, metabolic syndrome, marathon runners), a reduction in the intergroup variability of the insulin sensitivity was obtained. Moreover, the proposed index based on the oral glucose tolerance test (OGTT), which considers the glucose metabolism process and the hepatic and peripheral insulin sensitivity, showed stronger correlations with the Stumvoll method and lower intergroup variability than the fasting one.

Joint analysis of fasting and postprandial plasma glucose and insulin concentrations in Venezuelan women.

AIMS: Plasma glucose and insulin concentrations in fasting and postprandial reflect the metabolism of glucose by the human body and are useful in the diagnosis of metabolic diseases, such as diabetes and insulin resistance. In this work, these concentrations are jointly analyzed in Venezuelan women and 28 classes that better specify each metabolic condition are generated. MATERIALS AND METHODS: Each class comprises a combination of fasting and postprandial ranges of glucose and insulin concentrations defined in the literature as normal, impaired and diabetic. A hypothesis test was used to find statistically significant differences between the classes, and confidence intervals for age and glucose and insulin concentrations were defined for each class. RESULTS AND CONCLUSION: The process of deterioration of glucose metabolism advances with the age of the subject, more than half of the prediabetics have impaired glucose levels in fasting but normal in postprandial and normal insulin levels in fasting and postprandial, and one third of diabetics have diabetic glucose levels in fasting and postprandial and normal insulin levels in fasting and postprandial. This categorization of subjects would allow the application of a more specific treatment and the possibility of predicting the progress of the metabolic disorder.

Optimal data fusion for the improvement of QRS complex detection in multi-channel ECG recordings.

The automatic analysis of the electrocardiogram (ECG) begins, traditionally, with the detection of QRS complexes. Afterwards, useful information can be extracted from it, ranging from the estimation of the instantaneous heart rate to nonlinear heart rate variability analysis. A plethora of works have been published on this topic; consequently, there exist many QRS complex detectors with high-performance values. However, just a few detectors have been conceived that profit from the information contained in several ECG leads to provide a robust QRS complex detection. In this work, we explore the fusion of multi-channel ECG recordings QRS detections as a means to improve the detection performance. This paper presents a decentralized multi-channel QRS complex fusion scheme that optimally combines single-channel detections to produce a single detection signal. Using six different widely used QRS complex detectors on the MIT-BIH Arrhythmia and INCART databases, a reduction in false and missed detections was achieved with the proposed approach compared with the single-channel counterpart. Furthermore, our detection results are comparable with the performance of other multi-channel detectors found in the literature, showing, in turn, various advantages in scalability, adaptability, and simplicity in the system's implementationGraphical AbstractN QRS complex detectors simultaneously monitor N ECG channels. Once a detection occurs in a given channel, a 150 ms long window is opened to look for detections in other channels. Within this window, yn = + 1 if a QRS complex is detected and yn = - 1 otherwise. A coefficient α n, obtained during a training period and related to the detection performance in channel n, multiplies the detection signal yn, so that greater weights are assigned to ECG channels where single-channel detectors performed better. Finally, the binary detection decision (f ) is obtained from the comparison of the weighted sum of single-channel detections (z) with a fixed threshold (ß).

Cardiac autonomic modulation in response to a glucose stimulus.

This paper focuses on the effect of a sudden increase of plasma glucose concentration in the cardiac autonomic modulation using time-domain and frequency-domain heart rate variability (HRV) measures. Plasma glucose and insulin levels, measured each 30 min during an oral glucose tolerance test, and [Formula: see text] (mean of the RR interval), SDNN (standard deviation of normal-to-normal heartbeats), rMSSD (root-mean-square of successive differences between normal heartbeats), TP (total spectral power), LF and HF (power of the low- and high-frequency bands), LF norm and HF norm (LF and HF in normalized units), and LF/HF ratio of the HRV signal, obtained from 5-min-long ECG recordings during each phase of the test, were analyzed for subjects with the metabolic syndrome, marathon runners, and a control group. Results show that, after the glucose load, subjects with the metabolic syndrome experienced an increased sympathetic and decreased parasympathetic tone, which suggests an imbalance in cardiac autonomic modulation as a consequence of hyperglycemia and hyperinsulinemia. The significance of this study lies in the use of the ECG to assess the effects of a sudden increase in plasma glucose concentration on the cardiac autonomic modulation in subjects with different cardiovascular and metabolic conditions. Graphical Abstract Time-domain and frequency-domain heart rate variability measures are altered in subjects with different cardiovascular and metabolic conditions during an oral glucose tolerance test.

Artificial Neural Network for Laparoscopic Skills Classification Using Motion Signals from Apple Watch.

The acquisition of laparoscopic technical skills is constrained by the limited training opportunities and the necessity of having staff physicians on site to provide feedback to the trainees. In addition, the assessment tools used to measure trainees performance are not always sensitive enough to detect different levels of expertise. To address this problem, two Apple Watches worn on inexperienced subjects in laparoscopy were used to record their motion signals (attitude, rotation rate and acceleration) during multiple practices of the peg transfer task in a fundamentals of laparoscopic surgery (FLS) trainer box. This training process was carried out through a massed practice methodology (two hours of training), in which subjects were assessed following the guidelines of the FLS program. Subsequently, a series of metrics were estimated from the acquired motion signals and the Spearman's rank correlation coefficient was used to select the most statistically significant attributes. Then, a classification model based on artificial neural networks was trained, using these attributes as model inputs, to classify trainees according to their level of expertise into three classes: low, intermediate and high. Using this approach, an average classification performance of F1=86.11% was achieved on a test subset. This suggests that new technologies, such as smartwatches, can be used to complement surgical training by including motion-based metrics to improve current clinical education and offering a new source of feedback through objective assessment.

Atrial Fibrillation Detection in Short Single Lead ECG Recordings Using Wavelet Transform and Artificial Neural Networks.

Atrial fibrillation (AF) is a common health issue, not only in developed countries but also in developing ones. AF can lead to strokes, heart failures, and even death if it is not diagnosed and treated on time, therefore automatic detection of AF is an urgent need, particularly using Internet- connected devices that can alert healthcare services. Detection of AF typically involves the analysis of electrocardiogram (ECG) recordings, where P-waves that characterize the atrial activity are substituted with f-waves of variable amplitude and duration. In this paper, we used the discrete wavelet transform to decompose the ECG signal into detail and approximation coefficients with different time-frequency resolutions. Features extracted from ECG signals, RR interval time series and detail and approximation coefficients were used as inputs to an artificial neural network trained to identify four classes of heart rhythms: normal sinus rhythm (NSR), AF, other rhythms (OR) and noisy signals (NS). By performing a Monte Carlo 10- fold cross-validation of 10 iterations approach, average micro F1 scores of 83.64%, 61.61%, 56.88% and 53.88% to classify NSR, AF, OR and NS respectively, and average macro F1 of 64.00% were obtained on the publicly available training set of PhysioNet/Computing in Cardiology Challenge 2017. In addition, in a one-vs.-the-rest strategy, i.e., AF-vs-the-rest, averages sensitivity and specificity of 95.70% and 72.39% respectively were achieved to classify AF recordings.

Comparing glucose and insulin data from the two-hour oral glucose tolerance test in metabolic syndrome subjects and marathon runners.

Glucose is the main energy source of the body's cells and is essential for normal metabolism. Two pancreatic hormones, insulin and glucagon, are involved in glucose home-ostasis. Alteration in the plasma glucose and insulin concentrations could lead to distinct symptoms and diseases, ranging from mental function impairment to coma and even death. Type 2 diabetes, insulin resistance and metabolic syndrome are typical examples of abnormal glucose metabolism that increase the risk for cardiovascular disease and mortality. The oral glucose tolerance test (OGTT) is a medical test used to screen for prediabetes, type 2 diabetes and insulin resistance. In the 5-sample 2-hour OGTT, plasma glucose and insulin concentrations are measured after a fast and then after oral intake of glucose, at intervals of 30 minutes. In this work, a statistical analysis is carried out to find significant differences between the five stages of the OGTT for plasma glucose and insulin data. In addition, the behavior of the glucose and insulin data is compared between subjects with the metabolic syndrome and marathon runners. Results show that marathon runners have plasma glucose and insulin levels significantly lower (p <; 0.05) than people with the metabolic syndrome in all the stages of the OGTT. Insulin secretion decreases in marathon runners due to a significant reduction in plasma glucose concentration, but insulin secretion does not decrease in metabolic syndrome subjects due to insulin resistance, consequently plasma glucose concentration does not achieve normal levels.

Anthropometric measurements for assessing insulin sensitivity on patients with metabolic syndrome, sedentaries and marathoners.

The diagnosis of low insulin sensitivity is commonly done through the HOMA-IR index, in which fasting insulin and glucose blood levels are evaluated. Insulin and blood glucose levels are used for insulin sensitivity assessment by surrogate methods (HOMA-IR, Matsuda, etc), but anthropometric measurements like body weight, height and waist circumference are not considered, even if these variables also are related to low insulin sensitivity and metabolic syndrome. In this study we evaluate the impact of anthropometric measurements on the HOMA-IR, Matsuda and Caumo indexes to estimate insulin sensitivity. Specifically, we compare insulin sensitivity indexes with and without the anthropometric measurements in their equations on three different groups: patients with metabolic syndrome, sedentaries and marathoners. Results show relationships between anthropometric variables and insulin sensitivity indexes. On the other hand, subjects are mapped differently for insulin sensitivity assessment when anthropometric variables are taken into account. In addition, subjects diagnosed with normal insulin sensitivity could be considered as having low insulin sensitivity when anthropometric variables are considered.

Online apnea-bradycardia detection based on hidden semi-Markov models.

In this paper, we propose a new online apnea-bradycardia detection scheme that takes into account not only the instantaneous values of time series, but also their temporal evolution. The detector is based on a set of hidden semi-Markov models, representing the temporal evolution of beat-to-beat interval (RR interval) time series. A preprocessing step, including quantization and delayed version of the observation vector, is also proposed to maximize detection performance. This approach is quantitatively evaluated through simulated and real signals, the latter being acquired in neonatal intensive care units (NICU). Compared to two conventional detectors used in NICU, our best detector shows an improvement on average of around 15 % in sensitivity and 7 % in specificity. Furthermore, a reduced detection delay of approximately 2 s is also observed with respect to conventional detectors.

A new on-line electrocardiographic records database and computer routines for data analysis.

Gathering experimental data to test computer methods developed during a research is a hard work. Nowadays, some databases have been stored online that can be freely downloaded, however there is not a wide range of databases yet and not all pathologies are covered. Researchers with low resources are in need of more data they can consult for free. To cope with this we present an on-line portal containing a compilation of ECG databases recorded over the last two decades for research purposes. The first version of this portal contains four databases of ECG records: ischemic cardiopathy (72 patients, 3-lead ECG each), ischemic preconditioning (20 patients, 3-lead ECG each), diabetes (51 patients, 8-lead ECG each) and metabolic syndrome (25 subjects, 12-lead ECG each). In addition, one computer program and three routines are provided in order to correctly read the signals, and two digital filters along with two ECG waves detectors are provided for further processing. This portal will be constantly growing, other ECG databases and signal processing software will be uploaded. With this project, we give the scientific community a resource to avoid hours of data collection and to develop free software.

Adaptation of five indirect insulin sensitivity evaluation methods to three populations: metabolic syndrome, athletic and normal subjects.

Insulin sensitivity is determined using direct or indirect methods. Indirect methods are less invasive than direct methods, but have lower accuracy. The accuracy is set through the Spearman's rank correlation coefficient between the indirect method and a direct method. Since the set of parameters of each indirect method has been set empirically, different values of insulin sensitivity have been reported when they are applied on different populations. In this paper, five indirect methods (Avignon, HOMA-IR, QUICKI, Raynaud, and Matsuda) used to determine insulin sensitivity were adapted to three different populations: athletics, metabolic syndrome and normal subjects. The parameters of each method were varied in a range of values until the optimal value that gives the best correlation coefficient with a gold standard was obtained. Results show that the adaptation procedure led to an improved correlation coefficient. Additionally, the method of Matsuda was the most accurate, followed by the method of Avignon. We have confirmed that each indirect method needs a different set of parameters when it is applied to a specific population in order to obtain an accurate value of insulin sensitivity.

On-line apnea-bradycardia detection using hidden semi-Markov models.

In this work, we propose a detection method that exploits not only the instantaneous values, but also the intrinsic dynamics of the RR series, for the detection of apnea-bradycardia episodes in preterm infants. A hidden semi-Markov model is proposed to represent and characterize the temporal evolution of observed RR series and different pre-processing methods of these series are investigated. This approach is quantitatively evaluated through synthetic and real signals, the latter being acquired in neonatal intensive care units (NICU). Compared to two conventional detectors used in NICU our best detector shows an improvement of around 13% in sensitivity and 7% in specificity. Furthermore, a reduced detection delay of approximately 3 seconds is obtained with respect to conventional detectors.