Empirical comparison of deep learning models for fNIRS pain decoding.

Introduction: Pain assessment is extremely important in patients unable to communicate and it is often done by clinical judgement. However, assessing pain using observable indicators can be challenging for clinicians due to the subjective perceptions, individual differences in pain expression, and potential confounding factors. Therefore, the need for an objective pain assessment method that can assist medical practitioners. Functional near-infrared spectroscopy (fNIRS) has shown promising results to assess the neural function in response of nociception and pain. Previous studies have explored the use of machine learning with hand-crafted features in the assessment of pain. Methods: In this study, we aim to expand previous studies by exploring the use of deep learning models Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and (CNN-LSTM) to automatically extract features from fNIRS data and by comparing these with classical machine learning models using hand-crafted features. Results: The results showed that the deep learning models exhibited favourable results in the identification of different types of pain in our experiment using only fNIRS input data. The combination of CNN and LSTM in a hybrid model (CNN-LSTM) exhibited the highest performance (accuracy = 91.2%) in our problem setting. Statistical analysis using one-way ANOVA with Tukey's (post-hoc) test performed on accuracies showed that the deep learning models significantly improved accuracy performance as compared to the baseline models. Discussion: Overall, deep learning models showed their potential to learn features automatically without relying on manually-extracted features and the CNN-LSTM model could be used as a possible method of assessment of pain in non-verbal patients. Future research is needed to evaluate the generalisation of this method of pain assessment on independent populations and in real-life scenarios.

Multilevel Pain Assessment with Functional Near-Infrared Spectroscopy: Evaluating ΔHBO2 and ΔHHB Measures for Comprehensive Analysis.

Assessing pain in non-verbal patients is challenging, often depending on clinical judgment which can be unreliable due to fluctuations in vital signs caused by underlying medical conditions. To date, there is a notable absence of objective diagnostic tests to aid healthcare practitioners in pain assessment, especially affecting critically-ill or advanced dementia patients. Neurophysiological information, i.e., functional near-infrared spectroscopy (fNIRS) or electroencephalogram (EEG), unveils the brain's active regions and patterns, revealing the neural mechanisms behind the experience and processing of pain. This study focuses on assessing pain via the analysis of fNIRS signals combined with machine learning, utilising multiple fNIRS measures including oxygenated haemoglobin (ΔHBO2) and deoxygenated haemoglobin (ΔHHB). Initially, a channel selection process filters out highly contaminated channels with high-frequency and high-amplitude artifacts from the 24-channel fNIRS data. The remaining channels are then preprocessed by applying a low-pass filter and common average referencing to remove cardio-respiratory artifacts and common gain noise, respectively. Subsequently, the preprocessed channels are averaged to create a single time series vector for both ΔHBO2 and ΔHHB measures. From each measure, ten statistical features are extracted and fusion occurs at the feature level, resulting in a fused feature vector. The most relevant features, selected using the Minimum Redundancy Maximum Relevance method, are passed to a Support Vector Machines classifier. Using leave-one-subject-out cross validation, the system achieved an accuracy of 68.51%±9.02% in a multi-class task (No Pain, Low Pain, and High Pain) using a fusion of ΔHBO2 and ΔHHB. These two measures collectively demonstrated superior performance compared to when they were used independently. This study contributes to the pursuit of an objective pain assessment and proposes a potential biomarker for human pain using fNIRS.

"Where does it hurt?": Exploring EDA Signals to Detect and Localise Acute Pain.

Pain is a highly unpleasant sensory experience, for which currently no objective diagnostic test exists to measure it. Identification and localisation of pain, where the subject is unable to communicate, is a key step in enhancing therapeutic outcomes. Numerous studies have been conducted to categorise pain, but no reliable conclusion has been achieved. This is the first study that aims to show a strict relation between Electrodermal Activity (EDA) signal features and the presence of pain and to clarify the relation of classified signals to the location of the pain. For that purpose, EDA signals were recorded from 28 healthy subjects by inducing electrical pain at two anatomical locations (hand and forearm) of each subject. The EDA data were preprocessed with a Discrete Wavelet Transform to remove any irrelevant information. Chi-square feature selection was used to select features extracted from three domains: time, frequency, and cepstrum. The final feature vector was fed to a pool of classification schemes where an Artificial Neural Network classifier performed best. The proposed method, evaluated through leave-one-subject-out cross-validation, provided 90% accuracy in pain detection (no pain vs. pain), whereas the pain localisation experiment (hand pain vs. forearm pain) achieved 66.67% accuracy.Clinical relevance- This is the first study to provide an analysis of EDA signals in finding the source of the pain. This research explores the viability of using EDA for pain localisation, which may be helpful in the treatment of noncommunicable patients.

Multimodal physiological sensing for the assessment of acute pain.

Pain assessment is a challenging task encountered by clinicians. In clinical settings, patients' self-report is considered the gold standard in pain assessment. However, patients who are unable to self-report pain are at a higher risk of undiagnosed pain. In the present study, we explore the use of multiple sensing technologies to monitor physiological changes that can be used as a proxy for objective measurement of acute pain. Electrodermal activity (EDA), photoplethysmography (PPG), and respiration (RESP) signals were collected from 22 participants under two pain intensities (low and high) and on two different anatomical locations (forearm and hand). Three machine learning models were implemented, including support vector machines (SVM), decision trees (DT), and linear discriminant analysis (LDA) for the identification of pain. Various pain scenarios were investigated, identification of pain (no pain, pain), multiclass (no pain, low pain, high pain), and identification of pain location (forearm, hand). Reference classification results from individual sensors and from all sensors together were obtained. After feature selection, results showed that EDA was the most informative sensor in the three pain conditions, 93.2±8% in identification of pain, 68.9±10% in the multiclass problem, and 56.0±8% for the identification of pain location. These results identify EDA as the superior sensor in our experimental conditions. Future work is required to validate the obtained features to improve its feasibility in more realistic scenarios. Finally, this study proposes EDA as a candidate to design a tool that can assist clinicians in the assessment of acute pain of nonverbal patients.

A systematic review of neurophysiological sensing for the assessment of acute pain.

Pain is a complex and personal experience that presents diverse measurement challenges. Different sensing technologies can be used as a surrogate measure of pain to overcome these challenges. The objective of this review is to summarise and synthesise the published literature to: (a) identify relevant non-invasive physiological sensing technologies that can be used for the assessment of human pain, (b) describe the analytical tools used in artificial intelligence (AI) to decode pain data collected from sensing technologies, and (c) describe the main implications in the application of these technologies. A literature search was conducted in July 2022 to query PubMed, Web of Sciences, and Scopus. Papers published between January 2013 and July 2022 are considered. Forty-eight studies are included in this literature review. Two main sensing technologies (neurological and physiological) are identified in the literature. The sensing technologies and their modality (unimodal or multimodal) are presented. The literature provided numerous examples of how different analytical tools in AI have been applied to decode pain. This review identifies different non-invasive sensing technologies, their analytical tools, and the implications for their use. There are significant opportunities to leverage multimodal sensing and deep learning to improve accuracy of pain monitoring systems. This review also identifies the need for analyses and datasets that explore the inclusion of neural and physiological information together. Finally, challenges and opportunities for designing better systems for pain assessment are also presented.

Experimental Exploration of Multilevel Human Pain Assessment Using Blood Volume Pulse (BVP) Signals.

Critically ill patients often lack cognitive or communicative functions, making it challenging to assess their pain levels using self-reporting mechanisms. There is an urgent need for an accurate system that can assess pain levels without relying on patient-reported information. Blood volume pulse (BVP) is a relatively unexplored physiological measure with the potential to assess pain levels. This study aims to develop an accurate pain intensity classification system based on BVP signals through comprehensive experimental analysis. Twenty-two healthy subjects participated in the study, in which we analyzed the classification performance of BVP signals for various pain intensities using time, frequency, and morphological features through fourteen different machine learning classifiers. Three experiments were conducted using leave-one-subject-out cross-validation to better examine the hidden signatures of BVP signals for pain level classification. The results of the experiments showed that BVP signals combined with machine learning can provide an objective and quantitative evaluation of pain levels in clinical settings. Specifically, no pain and high pain BVP signals were classified with 96.6% accuracy, 100% sensitivity, and 91.6% specificity using a combination of time, frequency, and morphological features with artificial neural networks (ANNs). The classification of no pain and low pain BVP signals yielded 83.3% accuracy using a combination of time and morphological features with the AdaBoost classifier. Finally, the multi-class experiment, which classified no pain, low pain, and high pain, achieved 69% overall accuracy using a combination of time and morphological features with ANN. In conclusion, the experimental results suggest that BVP signals combined with machine learning can offer an objective and reliable assessment of pain levels in clinical settings.

Multimodal Fusion for Objective Assessment of Cognitive Workload: A Review.

Considerable progress has been made in improving the estimation accuracy of cognitive workload using various sensor technologies. However, the overall performance of different algorithms and methods remain suboptimal in real-world applications. Some studies in the literature demonstrate that a single modality is sufficient to estimate cognitive workload. These studies are limited to controlled settings, a scenario that is significantly different from the real world where data gets corrupted, interrupted, and delayed. In such situations, the use of multiple modalities is needed. Multimodal fusion approaches have been successful in other domains, such as wireless-sensor networks, in addressing single-sensor weaknesses and improving information quality/accuracy. These approaches are inherently more reliable when a data source is lost. In the cognitive workload literature, sensors, such as electroencephalography (EEG), electrocardiography (ECG), and eye tracking, have shown success in estimating the aspects of cognitive workload. Multimodal approaches that combine data from several sensors together can be more robust for real-time measurement of cognitive workload. In this article, we review the published studies related to multimodal data fusion to estimate the cognitive workload and synthesize their main findings. We identify the opportunities for designing better multimodal fusion systems for cognitive workload modeling.

Electroencephalographic Workload Indicators During Teleoperation of an Unmanned Aerial Vehicle Shepherding a Swarm of Unmanned Ground Vehicles in Contested Environments.

Background: Although many electroencephalographic (EEG) indicators have been proposed in the literature, it is unclear which of the power bands and various indices are best as indicators of mental workload. Spectral powers (Theta, Alpha, and Beta) and ratios (Beta/(Alpha + Theta), Theta/Alpha, Theta/Beta) were identified in the literature as prominent indicators of cognitive workload. Objective: The aim of the present study is to identify a set of EEG indicators that can be used for the objective assessment of cognitive workload in a multitasking setting and as a foundational step toward a human-autonomy augmented cognition system. Methods: The participants' perceived workload was modulated during a teleoperation task involving an unmanned aerial vehicle (UAV) shepherding a swarm of unmanned ground vehicles (UGVs). Three sources of data were recorded from sixteen participants (n = 16): heart rate (HR), EEG, and subjective indicators of the perceived workload using the Air Traffic Workload Input Technique (ATWIT). Results: The HR data predicted the scores from ATWIT. Nineteen common EEG features offered a discriminatory power of the four workload setups with high classification accuracy (82.23%), exhibiting a higher sensitivity than ATWIT and HR. Conclusion: The identified set of features represents EEG indicators for the objective assessment of cognitive workload across subjects. These common indicators could be used for augmented intelligence in human-autonomy teaming scenarios, and form the basis for our work on designing a closed-loop augmented cognition system for human-swarm teaming.

A Machine Learning Approach for the Identification of a Biomarker of Human Pain using fNIRS.

Pain is a highly unpleasant sensory and emotional experience, and no objective diagnosis test exists to assess it. In clinical practice there are two main methods for the estimation of pain, a patient's self-report and clinical judgement. However, these methods are highly subjective and the need of biomarkers to measure pain is important to improve pain management, reduce risk factors, and contribute to a more objective, valid, and reliable diagnosis. Therefore, in this study we propose the use of functional near-infrared spectroscopy (fNIRS) and machine learning for the identification of a possible biomarker of pain. We collected pain information from 18 volunteers using the thermal test of the quantitative sensory testing (QST) protocol, according to temperature level (cold and hot) and pain intensity (low and high). Feature extraction was completed in three different domains (time, frequency, and wavelet), and a total of 69 features were obtained. Feature selection was carried out according to three criteria, information gain (IG), joint mutual information (JMI), and Chi-squared (χ2). The significance of each feature ranking was evaluated using three learning models separately, linear discriminant analysis (LDA), the K-nearest neighbour (K-NN) and support vector machines (SVM) using the linear and Gaussian and polynomial kernels. The results showed that the Gaussian SVM presented the highest accuracy (94.17%) using only 25 features to identify the four types of pain in our database. In addition, we propose the use of the top 13 features according to the JMI criteria, which exhibited an accuracy of 89.44%, as promising biomarker of pain. This study contributes to the idea of developing an objective assessment of pain and proposes a potential biomarker of human pain using fNIRS.

Cortical Network Response to Acupuncture and the Effect of the Hegu Point: An fNIRS Study.

Acupuncture is a practice of treatment based on influencing specific points on the body by inserting needles. According to traditional Chinese medicine, the aim of acupuncture treatment for pain management is to use specific acupoints to relieve excess, activate qi (or vital energy), and improve blood circulation. In this context, the Hegu point is one of the most widely-used acupoints for this purpose, and it has been linked to having an analgesic effect. However, there exists considerable debate as to its scientific validity. In this pilot study, we aim to identify the functional connectivity related to the three main types of acupuncture manipulations and also identify an analgesic effect based on the hemodynamic response as measured by functional near-infrared spectroscopy (fNIRS). The cortical response of eleven healthy subjects was obtained using fNIRS during an acupuncture procedure. A multiscale analysis based on wavelet transform coherence was employed to assess the functional connectivity of corresponding channel pairs within the left and right somatosensory region. The wavelet analysis was focused on the very-low frequency oscillations (VLFO, 0.01⁻0.08 Hz) and the low frequency oscillations (LFO, 0.08⁻0.15 Hz). A mixed model analysis of variance was used to appraise statistical differences in the wavelet domain for the different acupuncture stimuli. The hemodynamic response after the acupuncture manipulations exhibited strong activations and distinctive cortical networks in each stimulus. The results of the statistical analysis showed significant differences ( p < 0.05 ) between the tasks in both frequency bands. These results suggest the existence of different stimuli-specific cortical networks in both frequency bands and the anaesthetic effect of the Hegu point as measured by fNIRS.

Physiological fluctuations show frequency-specific networks in fNIRS signals during resting state.

Physiological fluctuations are commonly present in functional studies of hemodynamic response such as functional magnetic resonance imaging (fMRI) and functional near-infrared spectroscopy (fNIRS). However, the effects of these signals in neural mechanisms are not fully understood. Thus, the aim of this study is to propose that frequency-specific networks exist in the somatosensory region within the frequency range of physiological fluctuations. We used a wavelet coherence approach to identify functional connectivity between cortical regions. Based on the spectral response, four frequency bands were identified: cardiac (0.8-1.5 Hz), respiration (0.16-0.6 Hz), low frequency oscillations (LFO) (0.04-0.15 Hz), and very low frequency oscillations (VLFO) (0.0130.04 Hz). Eight cortical networks were revealed after ipsilateral and contralateral analysis to evaluate connectivity in each frequency band. The ANOVA analysis proved the adequacy of the connectivity map for all frequencies bands. Finally, these findings suggest possible frequency-specific organizations within the frequency bands of physiological fluctuations in the resting human brain.

Toward a functional near-infrared spectroscopy-based monitoring of pain assessment for nonverbal patients.

Pain diagnosis for nonverbal patients represents a challenge in clinical settings. Neuroimaging methods, such as functional magnetic resonance imaging and functional near-infrared spectroscopy (fNIRS), have shown promising results to assess neuronal function in response to nociception and pain. Recent studies suggest that neuroimaging in conjunction with machine learning models can be used to predict different cognitive tasks. The aim of this study is to expand previous studies by exploring the classification of fNIRS signals (oxyhaemoglobin) according to temperature level (cold and hot) and corresponding pain intensity (low and high) using machine learning models. Toward this aim, we used the quantitative sensory testing to determine pain threshold and pain tolerance to cold and heat in 18 healthy subjects (three females), mean age±standard deviation (31.9±5.5). The classification model is based on the bag-of-words approach, a histogram representation used in document classification based on the frequencies of extracted words and adapted for time series; two learning algorithms were used separately, K-nearest neighbor (K-NN) and support vector machines (SVM). A comparison between two sets of fNIRS channels was also made in the classification task, all 24 channels and 8 channels from the somatosensory region defined as our region of interest (RoI). The results showed that K-NN obtained slightly better results (92.08%) than SVM (91.25%) using the 24 channels; however, the performance slightly dropped using only channels from the RoI with K-NN (91.53%) and SVM (90.83%). These results indicate potential applications of fNIRS in the development of a physiologically based diagnosis of human pain that would benefit vulnerable patients who cannot self-report pain.