Agnese Sbrollini

Segmented beat modulation method for electrocardiogram estimation from noisy recordings

Clinical utility of an electrocardiogram (ECG) affected by too high levels of noise such as baseline wanders, electrode motion artifacts, muscular artifacts and power-line interference may be jeopardized if not opportunely processed. Template-based techniques have been proposed for ECG estimation from noisy recordings, but usually they do not reproduce physiological ECG variability, which, however, provides clinically useful information on the patients health. Thus, this study proposes the Segmented-Beat Modulation Method (SBMM) as a new template-based filtering procedure able to reproduce ECG variability, and assesses SBMM robustness to the aforementioned noises in comparison to a standard template method (STM). SBMM performs a unique ECG segmentation into QRS segment and TUP segment, and successively modulates/demodulates (by stretching or compressing) the former segments in order to adaptively adjust each estimated beat to its original morphology and duration. Consequently, SBMM estimates ECG with significantly lower estimation errors than STM when applied to recordings affected by various levels of the considered noises (SBMM: 176-232µV and 79-499µV; STM: 215-496µV and 93-1056µV, for QRS and TUP segments, respectively). Thus, SBMM is able to reproduce ECG variability and is more robust to noise than STM.

Statistical baseline assessment in cardiotocography

Cardiotocography (CTG) is the most common non-invasive diagnostic technique to evaluate fetal well-being. It consists in the recording of fetal heart rate (FHR; bpm) and maternal uterine contractions. Among the main parameters characterizing FHR, baseline (BL) is fundamental to determine fetal hypoxia and distress. In computerized applications, BL is typically computed as mean FHR±ΔFHR, with ΔFHR=8 bpm or ΔFHR=10 bpm, both values being experimentally fixed. In this context, the present work aims: to propose a statistical procedure for ΔFHR assessment; to quantitatively determine ΔFHR value by applying such procedure to clinical data; and to compare the statistically-determined ΔFHR value against the experimentally-determined ΔFHR values. To these aims, the 552 recordings of the “CTU-UHB intrapartum CTG database” from Physionet were submitted to an automatic procedure, which consisted in a FHR preprocessing phase and a statistical BL assessment. During preprocessing, FHR time series were divided into 20-min sliding windows, in which missing data were removed by linear interpolation. Only windows with a correction rate lower than 10% were further processed for BL assessment, according to which ΔFHR was computed as FHR standard deviation. Total number of accepted windows was 1192 (38.5%) over 383 recordings (69.4%) with at least an accepted window. Statistically-determined ΔFHR value was 9.7 bpm. Such value was statistically different from 8 bpm (P<10−19) but not from 10 bpm (P=0.16). Thus, ΔFHR=10 bpm is preferable over 8 bpm because both experimentally and statistically validated.

Heart Rate Detection Using Microsoft Kinect: Validation and Comparison to Wearable Devices

Contactless detection is one of the new frontiers of technological innovation in the field of healthcare, enabling unobtrusive measurements of biomedical parameters. Compared to conventional methods for Heart Rate (HR) detection that employ expensive and/or uncomfortable devices, such as the Electrocardiograph (ECG) or pulse oximeter, contactless HR detection offers fast and continuous monitoring of heart activities and provides support for clinical analysis without the need for the user to wear a device. This paper presents a validation study for a contactless HR estimation method exploiting RGB (Red, Green, Blue) data from a Microsoft Kinect v2 device. This method, based on Eulerian Video Magnification (EVM), Photoplethysmography (PPG) and Videoplethysmography (VPG), can achieve performance comparable to classical approaches exploiting wearable systems, under specific test conditions. The output given by a Holter, which represents the gold-standard device used in the test for ECG extraction, is considered as the ground-truth, while a comparison with a commercial smartwatch is also included. The validation process is conducted with two modalities that differ for the availability of a priori knowledge about the subjects’ normal HR. The two test modalities provide different results. In particular, the HR estimation differs from the ground-truth by 2% when the knowledge about the subject’s lifestyle and his/her HR is considered and by 3.4% if no information about the person is taken into account.

Noninvasive Fetal Electrocardiography Part I: Pan-Tompkins' Algorithm Adaptation to Fetal R-peak Identification

Background: Indirect fetal electrocardiography is preferable to direct fetal electrocardiography because of being noninvasive and is applicable also during the end of pregnancy, besides labor. Still, the former is strongly affected by noise so that even R-peak detection (which is essential for fetal heart-rate evaluations and subsequent processing procedures) is challenging. Some fetal studies have applied the Pan-Tompkins’ algorithm that, however, was originally designed for adult applications. Thus, this work evaluated the Pan-Tompkins’ algorithm suitability for fetal applications, and proposed fetal adjustments and optimizations to improve it. Method: Both Pan-Tompkins’ algorithm and its improved version were applied to the “Abdominal and Direct Fetal Electrocardiogram Database” and to the “Noninvasive Fetal Electrocardiography Database” of Physionet. R-peak detection accuracy was quantified by computation of positive-predictive value, sensitivity and F1 score. Results: When applied to “Abdominal and Direct Fetal Electrocardiogram Database”, the accuracy of the improved fetal Pan-Tompkins’ algorithm was significantly higher than the standard (positive-predictive value: 0.94 vs. 0.79; sensitivity: 0.95 vs. 0.80; F1 score: 0.94 vs. 0.79; P<0.05 in all cases) on indirect fetal electrocardiograms, whereas both methods performed similarly on direct fetal electrocardiograms (positive-predictive value, sensitivity and F1 score all close to 1). Improved fetal Pan-Tompkins’ algorithm was found to be superior to the standard also when applied to “Noninvasive Fetal Electrocardiography Database” (positive-predictive value: 0.68 vs. 0.55, P<0.05; sensitivity: 0.56 vs. 0.46, P=0.23; F1 score: 0.60 vs. 0.47, P=0.11). Conclusion: In indirect fetal electrocardiographic applications, improved fetal Pan-Tompkins’ algorithm is to be preferred over the standard, since it provides higher R-peak detection accuracy for heart-rate evaluations and subsequent processing.

Noninvasive Fetal Electrocardiography Part II: Segmented-Beat Modulation Method for Signal Denoising

Background: Fetal well-being evaluation may be accomplished by monitoring cardiac activity through fetal electrocardiography. Direct fetal electrocardiography (acquired through scalp electrodes) is the gold standard but its invasiveness limits its clinical applicability. Instead, clinical use of indirect fetal electrocardiography (acquired through abdominal electrodes) is limited by its poor signal quality. Objective: Aim of this study was to evaluate the suitability of the Segmented-Beat Modulation Method to denoise indirect fetal electrocardiograms in order to achieve a signal-quality at least comparable to the direct ones. Method: Direct and indirect recordings, simultaneously acquired from 5 pregnant women during labor, were filtered with the Segmented-Beat Modulation Method and correlated in order to assess their morphological correspondence. Signal-to-noise ratio was used to quantify their quality. Results: Amplitude was higher in direct than indirect fetal electrocardiograms (median:104 µV vs. 22 µV; P=7.66·10-4), whereas noise was comparable (median:70 µV vs. 49 µV, P=0.45). Moreover, fetal electrocardiogram amplitude was significantly higher than affecting noise in direct recording (P=3.17·10-2) and significantly in indirect recording (P=1.90·10-3). Consequently, signal-to-noise ratio was initially higher for direct than indirect recordings (median:3.3 dB vs. -2.3 dB; P=3.90·10-3), but became lower after denoising of indirect ones (median:9.6 dB; P=9.84·10-4). Eventually, direct and indirect recordings were highly correlated (median: ρ=0.78; P<10-208), indicating that the two electrocardiograms were morphologically equivalent. Conclusion: Segmented-Beat Modulation Method is particularly useful for denoising of indirect fetal electrocardiogram and may contribute to the spread of this noninvasive technique in the clinical practice.

Separation of Superimposed Electrocardiographic and Electromyographic Signals

Electrocardiography (ECG) and surface electromyography (SEMG) are two non-invasive tests to evaluate cardiac and muscular functionality, respectively. They are both acquired by placing electrodes on the body surface so they become one the interference of the other. Typically, linear filters are used for ECG and SEMG separation: high-pass filters with cutoff at 20 Hz to attenuate ECG interference in SEMG, and low-pass filters with cut-off at 50 Hz to attenuate SEMG interference in ECG. In spite of that, linear filtering is not adequate due to the presence of a 20-50 Hz frequency-band in which the two signal spectra overlap. The aim of the present study was to evaluate the ability of the Segmented-Beat Modulation Method (SBMM) for ECG and SEMG separation and by accurately maintaining signals characteristics. SBMM is a template-based technique for ECG denoising: under the hypothesis of ECG and SEMG linearly superimposed, it first provides an ECG estimation, and then an SEMG estimation by subtraction. In order to test the method under several conditions, SBMM was applied to simulated as well as clinical recordings with superimposed ECG and SEMG. SBMM was able to accurately estimate both ECG and SEMG in all cases. Indeed, ECG and SEMG were estimated by maintain their features such as amplitude (estimation errors <6%), heart rate and heart-rate variability. Moreover, estimated ECG was always characterized by a spectrum mostly (76.4-100.0%) included in the 0-50 Hz frequency-band, whereas estimated SEMG was always characterized by a spectrum mostly (80.9-95.6%) included in the 20-450 Hz frequency-band. Such results confirm the existence of a 20-50 Hz frequency-band in which ECG and SEMG spectral components are overlapped. Thus, SBMM is a robust filtering procedure to separate superimposed ECG and SEMG.

Evaluation of the low-frequency components in surface electromyography

The surface electromyogram (SEMG) is a signal noninvasively (through electrodes located on the body surface) acquired for evaluating the electrical activity produced by skeletal muscles. In thoracic acquisitions, SEMG is typically affected by the electrocardiographic (ECG) signal, representing the electrical activity of the heart. SEMG and ECG power spectra mainly fall within 5-450 Hz and 0.05-50 Hz, respectively. Consequently, SEMG and ECG components overlap in the 5-50 Hz range. Although removing SEMG low spectral components, high-pass linear filtering (LF) with a cut-off frequency of 20 Hz remains the standard technique to clean SEMG from ECG. Thus, the aim of the present study was to propose the Segmented Beat Modulation Method (SBMM) as a tool to clean SEMG from ECG without removing SEMG components below 20 Hz. A SEMG recording was first acquired from the left rectus abdominis of a subject, and then filtered using both SBMM and LF. Filtered SEMGs obtained with the two techniques were compared. Results indicate that SBMM eliminates ECG interference from SEMG better than LF, since the latter procedure maintains ECG components between 20 and 50 Hz. In addition, after ECG removal by SBMM, SEMG showed a significant amount of spectral components (up to 20%) in the low-frequency range. Maintaining such low-frequency components, which go lost when applying LF, is desirable since they likely associate to the firing rates of the active motor units. In conclusion, SBMM represents a useful tool to clean SEMG from ECG without loss of frequency components.

Association between Accelerations and Decelerations of Fetal Heart Rate

Cardiotocography (CTG) is the most popular test for establishing the fetal health status. Among its characterizing features there are the fetal heart rate (FHR) accelerations (ACC), usually considered a sign of fetal well-being; and decelerations (DEC), some of which may indicate the risk of fetal hypoxia. Thus, ACC and DEC are usually considered independent phenomena possibly providing opposite information on the fetus clinical status. CTG is typically analyzed by visual inspection; still a computerized analysis may provide a more objective CTG interpretation and precise ACC and DEC characterization. Aim of the present study is to propose an automatic procedure for ACC and DEC identification and characterization, and to investigate a potential relationship between their occurrence. The 552 tracings of the Physionet “CTU-CHB intrapartum CTG database” were analyzed according to a procedure that includes: FHR pre-processing; 20 min windowing; baseline estimation; and ACC and DEC identification and characterization. Specifically, ACC and DEC were defined as FHR deviations from baseline of at least 15 bpm for at least 15 s and then characterized in terms of length (s), amplitude (bpm) and area (length.amplitude; bpm.s). Only 383 (69.4%) CTG recordings showed sufficiently good FHR signal quality to be enrolled in the study. Number of DEC per window was significantly higher than ACC (4.0 vs 2.5; P<10-14). DEC were characterized by a comparable length but high-er amplitude and area than ACC (LNG: 56 s vs 61 s, P=0.2573; AMP: 12 bpm vs 10 bpm, P<10-11; AREA: 688 s·bpm vs 618 s·bpm, P=0.0032). DEC total area in a 20-min window was higher than that of ACC (3074 s·bpm vs 2007 s·bpm, P<10-9), but such areas were also strictly correlated (ρ=0.72; P<10-62). Thus, in a CTG recording, ACC and DEC are not independent phenomena but their occurrence is strictly associated.

Co-activation periods of gastrocnemius and vastus lateralis during walking evaluated by surface electromyography

“In vivo” studies reported that the co-activation of gastrocnemius and quadriceps femoris (QF) muscles produces ACL strain values greater than those caused by an isolated activation of either muscle. Aim of this study was to assess the co-activation of gastrocnemius (lateral head, GL) and vastus lateralis (VL) in healthy and young adults during walking. To this purpose the Statistical Gait Analysis was performed, that allows a characterization of gait considering hundreds of strides belonging to the same walking trial. Three GL/VL co-activations were detected during a single gait cycle: in foot-contact phase, from 6.8±8.5% to 22.9±23.3% of gait cycle, (FC co-activation), in push-off phase, from 33.0±11.9% to 41.5±13.4% (PO co-activation), and in swing phase, from 86.5±6.7% to 93.2±5.9% (SW co-activation). FC co-activation was the most recurrent (in 100% of the strides, P<;0.001) and longest (16% of gait cycle) one. Thus, the ACL strain due to the co-activation between GL and VL is longer and more frequently during FC phase, than in all the others gait phases. Moreover, the position of the knee and the amount of the weight-bearing on knee, achieved in this gait phase, suggested that FC co-activation is the one that produces a highest strain value of anterior cruciate ligament (ACL). These findings could help to better understand risk factors of the ACL injuries and to design more focused preventive and rehabilitative strategies.