-
Contactless pulse rate assessment: Results and insights for application in driving simulator
Authors:
Đorđe D. Nešković,
Kristina Stojmenova Pečečnik,
Jaka Sodnik,
Nadica Miljković
Abstract:
Remote Photoplethysmography (rPPG) enables continuous and unobtrusive assessment of driver's state, allowing estimation of fatigue, stress, and user experience. Commonly used wearable PPG sensors, while effective, suffer from motion artifacts and user discomfort. This study explores the feasibility of non-contact Pulse Rate (PR) assessment using facial video recordings captured by a Red, Green, an…
▽ More
Remote Photoplethysmography (rPPG) enables continuous and unobtrusive assessment of driver's state, allowing estimation of fatigue, stress, and user experience. Commonly used wearable PPG sensors, while effective, suffer from motion artifacts and user discomfort. This study explores the feasibility of non-contact Pulse Rate (PR) assessment using facial video recordings captured by a Red, Green, and Blue (RGB) camera in a driving simulator. The proposed approach detects subtle skin color variations due to blood flow and compares extracted PR values against reference measurements from a wearable Empatica E4 sensor. We evaluate the impact of Eulerian Video Magnification (EVM) on PR estimation and assess statistical differences in PR between age groups. Data obtained from 80 recordings from 64 healthy subjects covering a PR range of 45-160 bpm are analyzed, and PR extraction accuracy is quantified using metrics, such as Mean Absolute Error (MAE). Results show that EVM slightly improves PR estimation, reducing MAE from 6.48 bpm to 5.04 bpm. Significant differences between PRs in younger and older drivers are observed in both rPPG and reference data. The study also thoroughly discusses bias within Empatica E4. Overall, results support the feasibility of integrating camera-based PR monitoring into driving simulators for real-time subject's assessment.
△ Less
Submitted 27 June, 2025; v1 submitted 2 May, 2025;
originally announced May 2025.
-
Robotic Trail Maker Platform for Rehabilitation in Neurological Conditions: Clinical Use Cases
Authors:
Srikar Annamraju,
Harris Nisar,
Dayu Xia,
Shankar A. Deka,
Anne Horowitz,
Nadica Miljković,
Dušan M. Stipanović
Abstract:
Patients with neurological conditions require rehabilitation to restore their motor, visual, and cognitive abilities. To meet the shortage of therapists and reduce their workload, a robotic rehabilitation platform involving the clinical trail making test is proposed. Therapists can create custom trails for each patient and the patient can trace the trails using a robotic device. The platform can t…
▽ More
Patients with neurological conditions require rehabilitation to restore their motor, visual, and cognitive abilities. To meet the shortage of therapists and reduce their workload, a robotic rehabilitation platform involving the clinical trail making test is proposed. Therapists can create custom trails for each patient and the patient can trace the trails using a robotic device. The platform can track the performance of the patient and use these data to provide dynamic assistance through the robot to the patient interface. Therefore, the proposed platform not only functions as an evaluation platform, but also trains the patient in recovery. The developed platform has been validated at a rehabilitation center, with therapists and patients operating the device. It was found that patients performed poorly while using the platform compared to healthy subjects and that the assistance provided also improved performance amongst patients. Statistical analysis demonstrated that the speed of the patients was significantly enhanced with the robotic assistance. Further, neural networks are trained to classify between patients and healthy subjects and to forecast their movements using the data collected.
△ Less
Submitted 30 April, 2025; v1 submitted 27 April, 2025;
originally announced April 2025.
-
Data Augmentation for Generating Synthetic Electrogastrogram Time Series
Authors:
Nadica Miljković,
Nikola Milenić,
Nenad B. Popović,
Jaka Sodnik
Abstract:
To address an emerging need for large number of diverse datasets for rigor evaluation of signal processing techniques, we developed and evaluated a new method for generating synthetic electrogastrogram time series. We used electrogastrography (EGG) data from an open database to set model parameters and statistical tests to evaluate synthesized data. Additionally, we illustrated method customizatio…
▽ More
To address an emerging need for large number of diverse datasets for rigor evaluation of signal processing techniques, we developed and evaluated a new method for generating synthetic electrogastrogram time series. We used electrogastrography (EGG) data from an open database to set model parameters and statistical tests to evaluate synthesized data. Additionally, we illustrated method customization for generating artificial EGG time series alterations caused by the simulator sickness. Proposed data augmentation method generates synthetic EGG data with specified duration, sampling frequency, recording state (postprandial or fasting state), overall noise and breathing artifact injection, and pauses in the gastric rhythm (arrhythmia occurrence) with statistically significant difference between postprandial and fasting states in > 70% cases while not accounting for individual differences. Features obtained from the synthetic EGG signal resembling simulator sickness occurrence displayed expected trends. The code for generation of synthetic EGG time series is not only freely available and can be further customized to assess signal processing algorithms but also may be used to increase data diversity for training artificial intelligence (AI) algorithms. The proposed approach is customized for EGG data synthesis but can be easily utilized for other biosignals with similar nature such as electroencephalogram.
△ Less
Submitted 8 May, 2024; v1 submitted 4 March, 2023;
originally announced March 2023.
-
A New Weighted Time Window-based Method to Detect B-point in Impedance Cardiogram
Authors:
Nadica Miljković,
Tomislav B. Šekara
Abstract:
Background: Impedance Cardiogram (ICG) has impressive number of applications in healthcare. However, its wider adoption is excessively limited due to the well recognized challenges in ICG delineation. We present a simple, adaptive, and efficient method for the most demanding ICG delineation task - detection of less distinct B-point that marks the onset of the left ventricular ejection. Method: The…
▽ More
Background: Impedance Cardiogram (ICG) has impressive number of applications in healthcare. However, its wider adoption is excessively limited due to the well recognized challenges in ICG delineation. We present a simple, adaptive, and efficient method for the most demanding ICG delineation task - detection of less distinct B-point that marks the onset of the left ventricular ejection. Method: The core of the new method is transformation of ICG time series by weighted time window of an ICG segment preceding the maximal ICG peak (the C-point) aiming at the B-point enhancement. The resulting Modified B-point (MB-point) is then easily delineated. To evaluate the proposed workflow, the dataset comprising 20 healthy participants and 21065 B-points are manually annotated and openly shared with the software code. To the best of our knowledge, our ICG dataset has the largest number of annotated B-points. Detector performance is evaluated on two recorded segments with distinct noises, as well as on an open dataset comprising ICG recorded in another set of 24 healthy subjects. Results: The results showed that the method was superior to the competing methods when the tolerance for B-point detection was set to +/-150 ms in all cases and for both datasets (>99.4%). Conclusions: Proposed approach based on the weighted time windows presents a promising technique for reliable ICG delineation that is instrumental for appropriate feature extraction and ICG characterization. Further customization could incorporate labeling of other pathological ICG segments, other biomedical signals, as well as additional enhancements for improving robustness to noise.
△ Less
Submitted 24 May, 2023; v1 submitted 10 July, 2022;
originally announced July 2022.
-
Effectiveness of a time to fixate for fitness to drive evaluation in neurological patients
Authors:
Nadica Miljković,
Jaka Sodnik
Abstract:
We present a method to automatically calculate time to fixate (TTF) from the eye-tracker data in subjects with neurological impairment using a driving simulator. TTF presents the time interval for a person to notice the stimulus from its first occurrence. Precisely, we measured the time since the children started to cross the street until the drivers directed their look to the children. From 108 n…
▽ More
We present a method to automatically calculate time to fixate (TTF) from the eye-tracker data in subjects with neurological impairment using a driving simulator. TTF presents the time interval for a person to notice the stimulus from its first occurrence. Precisely, we measured the time since the children started to cross the street until the drivers directed their look to the children. From 108 neurological patients recruited for the study, the analysis of TTF was performed in 56 patients to assess fit-, unfit-, and conditionally-fit-to-drive patients. The results showed that the proposed method based on the YOLO (you only look once) object detector is efficient for computing TTFs from the eye-tracker data. We obtained discriminative results for fit-to-drive patients by application of Tukey's honest significant difference post hoc test (p < 0.01), while no difference was observed between conditionally-fit and unfit-to-drive groups (p = 0.542). Moreover, we show that time-to-collision (TTC), initial gaze distance (IGD) from pedestrians, and speed at the hazard onset did not influence the result, while the only significant interaction is among fitness, IGD, and TTC on TTF. Obtained TTFs are also compared with the perception response times (PRT) calculated independently from eye-tracker data and YOLO. Although we reached statistically significant results that speak in favor of possible method application for assessment of fitness to drive, we provide detailed directions for future driving simulation-based evaluation and propose processing workflow to secure reliable TTF calculation and its possible application in for example psychology and neuroscience.
△ Less
Submitted 25 July, 2023; v1 submitted 18 May, 2022;
originally announced May 2022.
-
Combined approach for automatic and robust calculation of dominant frequency of electrogastrogram
Authors:
Nebojša Jovanović,
Nenad B. Popović,
Nadica Miljković
Abstract:
We present a novel method for automatic and robust detection of dominant frequency (DF) in the electrogastrogram (EGG). Our new approach combines Fast Fourier Transform (FFT), Welch's method for spectral density estimation, and autocorrelation. The proposed combined method as well as other separate procedures were tested on a freely available dataset consisted of EGG recordings in 20 healthy indiv…
▽ More
We present a novel method for automatic and robust detection of dominant frequency (DF) in the electrogastrogram (EGG). Our new approach combines Fast Fourier Transform (FFT), Welch's method for spectral density estimation, and autocorrelation. The proposed combined method as well as other separate procedures were tested on a freely available dataset consisted of EGG recordings in 20 healthy individuals. DF was calculated in relation (1) to the fasting and postprandial states, (2) to the three recording locations, and (3) to the subjects' body mass index. For the estimation of algorithms performance in the presence of noise, we created a synthetic dataset by adding white Gaussian noise to the artifact-free EGG waveform in one subject. The individual algorithms and novel combined approach were evaluated in relation to the signal-to-noise ratio (SNR) in range from -40 dB to 20 dB. Our results showed that the novel combined method significantly outperformed the commonly used approach for DF calculation - FFT in noise presence when compared to the benchmark data being was manually corrected by an expert. The novel method outperformed autocorrelation and Welch's method in accuracy. Additionally, we presented a method for optimal window width selection when using Welch's spectrogram that showed that for DF detection, window length of N/4 (300 s), where N is the length of EGG waveform in samples, performed the best when compared to the benchmark data. The combined approach proved efficient for automatic and robust calculation of dominant frequency on openly available EGG dataset recorded in healthy individuals and is promising approach for DF detection.
△ Less
Submitted 18 September, 2020;
originally announced September 2020.
-
Effect of the sEMG electrode (re)placement and feature set size on the hand movement recognition
Authors:
Nadica Miljković,
Milica S. Isaković
Abstract:
Repositioning of recording electrode array across repeated electromyography measurements may result in a displacement error in hand movement classification systems. In order to examine if the classifier re-training could reach satisfactory results when electrode array is translated along or rotated around subject's forearm for varying number of features, we recorded surface electromyography signal…
▽ More
Repositioning of recording electrode array across repeated electromyography measurements may result in a displacement error in hand movement classification systems. In order to examine if the classifier re-training could reach satisfactory results when electrode array is translated along or rotated around subject's forearm for varying number of features, we recorded surface electromyography signals in 10 healthy volunteers for three types of grasp and 6 wrist movements. For feature extraction we applied principal component analysis and the feature set size varied from one to 8 principal components. We compared results of re-trained classifier with results from leave-one-out cross-validation classification procedure for three classifiers: LDA (Linear Discriminant Analysis), QDA (Quadratic Discriminant Analysis), and ANN (Artificial Neural Network). Our results showed that there was no significant difference in classification accuracy when the array electrode was repositioned indicating successful classification re-training and optimal feature set selection. The results also indicate expectedly that the number of principal components plays a key role for acceptable classification accuracy ~90 %. For the largest dataset (9 hand movements), LDA and QDA outperformed ANN, while for three grasping movements ANN showed promising results. Interestingly, we showed that interaction between electrode array position and the feature set size is not statistically significant. This study emphasizes the importance of testing the interaction of factors that influence classification accuracy and classifier selection altogether with their impact independently in order to establish guiding principles for design of hand movement recognition system. Data recorded for this study are stored on Zenodo repository (doi: 10.5281/zenodo.4039550).
△ Less
Submitted 17 November, 2020; v1 submitted 5 May, 2020;
originally announced May 2020.
