Tonic-clonic seizure detection using accelerometry-based wearable sensors: A prospective, video-EEG controlled study.

PURPOSE: The aim of this prospective, video-electroencephalography (video-EEG) controlled study was to evaluate the performance of an accelerometry-based wearable system to detect tonic-clonic seizures (TCSs) and to investigate the accuracy of different seizure detection algorithms using separate training and test data sets. METHODS: Seventy-five epilepsy surgery candidates undergoing video-EEG monitoring were included. The patients wore one three-axis accelerometer on each wrist during video-EEG. The accelerometer data was band-pass filtered and reduced using a movement threshold and mapped to a time-frequency feature space representation. Algorithms based on standard binary classifiers combined with a TCS specific event detection layer were developed and trained using the training set. Their performance was evaluated in terms of sensitivity and false positive (FP) rate using the test set. RESULTS: Thirty-seven available TCSs in 11 patients were recorded and the data was divided into disjoint training (27 TCSs, three patients) and test (10 TCSs, eight patients) data sets. The classification algorithms evaluated were K-nearest-neighbors (KNN), random forest (RF) and a linear kernel support vector machine (SVM). For the TCSs detection performance of the three algorithms in the test set, the highest sensitivity was obtained for KNN (100% sensitivity, 0.05 FP/h) and the lowest FP rate was obtained for RF (90% sensitivity, 0.01 FP/h). CONCLUSIONS: The low FP rate enhances the clinical utility of the detection system for long-term reliable seizure monitoring. It also allows a possible implementation of an automated TCS detection in free-living environment, which could contribute to ascertain seizure frequency and thereby better seizure management.

CLAss-Specific Subspace Kernel Representations and Adaptive Margin Slack Minimization for Large Scale Classification.

In kernel-based classification models, given limited computational power and storage capacity, operations over the full kernel matrix becomes prohibitive. In this paper, we propose a new supervised learning framework using kernel models for sequential data processing. The framework is based on two components that both aim at enhancing the classification capability with a subset selection scheme. The first part is a subspace projection technique in the reproducing kernel Hilbert space using a CLAss-specific Subspace Kernel representation for kernel approximation. In the second part, we propose a novel structural risk minimization algorithm called the adaptive margin slack minimization to iteratively improve the classification accuracy by an adaptive data selection. We motivate each part separately, and then integrate them into learning frameworks for large scale data. We propose two such frameworks: the memory efficient sequential processing for sequential data processing and the parallelized sequential processing for distributed computing with sequential data acquisition. We test our methods on several benchmark data sets and compared with the state-of-the-art techniques to verify the validity of the proposed techniques.

The effects of ventricular drainage on the intracranial pressure signal and the pressure reactivity index.

In subarachnoid hemorrhage (SAH) patients intracranial pressure (ICP) is usually monitored via an extraventricular drain (EVD), which can produce false readings when the drain is open. It is established that both the ICP cardiac pulse frequency and long term trends over several hours are often seriously corrupted. The aim of this study was to establish whether or not the intermediate frequency bands [respiratory, Mayer wave and very low frequency (VLF)] were also corrupted. The VLF range is of special interest because it is important in cerebral autoregulation studies. Using a pattern recognition algorithm we retrospectively identified 718 cases of EVD opening in 80 SAH patients. An analysis of differences between closed and open-drain periods showed that ICP amplitude decreased significantly in all of the three lower frequency bands when the EVD was open. A similar analysis of systemic arterial pressure signal revealed similar changes in the same frequency bands that were positively correlated with the ICP changes. Therefore we concluded that the changes in the ICP signal represented real, physiological changes and not artifact. Pressure reactivity index (PRx) values were also computed during closed and open-drain periods. We found a small but statistically significant decrease during open-drain periods. Based on analysis of the change in the PRx distribution during open drainage we concluded that this decrease also represented physiological changes rather than artifact. In summary the ICP respiratory, Mayer wave, and VLF frequency bands are not corrupted when the EVD is open, and it safe to use these for autoregulation studies.

Microwave technology for detecting traumatic intracranial bleedings: tests on phantom of subdural hematoma and numerical simulations.

Traumatic brain injury is the leading cause of death and severe disability for young people and a major public health problem for elderly. Many patients with intracranial bleeding are treated too late, because they initially show no symptoms of severe injury and are not transported to a trauma center. There is a need for a method to detect intracranial bleedings in the prehospital setting. In this study, we investigate whether broadband microwave technology (MWT) in conjunction with a diagnostic algorithm can detect subdural hematoma (SDH). A human cranium phantom and numerical simulations of SDH are used. Four phantoms with SDH 0, 40, 70 and 110 mL are measured with a MWT instrument. The simulated dataset consists of 1500 observations. Classification accuracy is assessed using fivefold cross-validation, and a validation dataset never used for training. The total accuracy is 100 and 82-96 % for phantom measurements and simulated data, respectively. Sensitivity and specificity for bleeding detection were 100 and 96 %, respectively, for the simulated data. SDH of different sizes is differentiated. The classifier requires training dataset size in order of 150 observations per class to achieve high accuracy. We conclude that the results indicate that MWT can detect and estimate the size of SDH. This is promising for developing MWT to be used for prehospital diagnosis of intracranial bleedings.

An optimal frequency range for assessing the pressure reactivity index in patients with traumatic brain injury.

The objective of this study was to identify the optimal frequency range for computing the pressure reactivity index (PRx). PRx is a clinical method for assessing cerebral pressure autoregulation based on the correlation of spontaneous variations of arterial blood pressure (ABP) and intracranial pressure (ICP). Our hypothesis was that optimizing the methodology for computing PRx in this way could produce a more stable, reliable and clinically useful index of autoregulation status. The patients studied were a series of 131 traumatic brain injury patients. Pressure reactivity indices were computed in various frequency bands during the first 4 days following injury using bandpass filtering of the input ABP and ICP signals. Patient outcome was assessed using the extended Glasgow Outcome Scale (GOSe). The optimization criterion was the strength of the correlation with GOSe of the mean index value over the first 4 days following injury. Stability of the indices was measured as the mean absolute deviation of the minute by minute index value from 30-min moving averages. The optimal index frequency range for prediction of outcome was identified as 0.018-0.067 Hz (oscillations with periods from 55 to 15 s). The index based on this frequency range correlated with GOSe with ρ=-0.46 compared to -0.41 for standard PRx, and reduced the 30-min variation by 23%.

Microwave-based stroke diagnosis making global prehospital thrombolytic treatment possible.

Here, we present two different brain diagnostic devices based on microwave technology and the associated two first proof-of-principle measurements that show that the systems can differentiate hemorrhagic from ischemic stroke in acute stroke patients, as well as differentiate hemorrhagic patients from healthy volunteers. The system was based on microwave scattering measurements with an antenna system worn on the head. Measurement data were analyzed with a machine-learning algorithm that is based on training using data from patients with a known condition. Computer tomography images were used as reference. The detection methodology was evaluated with the leave-one-out validation method combined with a Monte Carlo-based bootstrap step. The clinical motivation for this project is that ischemic stroke patients may receive acute thrombolytic treatment at hospitals, dramatically reducing or abolishing symptoms. A microwave system is suitable for prehospital use, and therefore has the potential to allow significantly earlier diagnosis and treatment than today.

Microwave technology for localization of traumatic intracranial bleedings-a numerical simulation study.

Traumatic brain injury (TBI) is a major public health problem worldwide. Intracranial bleedings represents the most serious complication of TBI and need to be surgically evacuated promptly to save lives and mitigate injury. Microwave technology (MWT) is promising as a complement to computed tomography (CT) to be used in road and air ambulances for early detection of intracranial bleedings. In this study, we perform numerical simulations to investigate if a classification algorithm based on singular value decomposition can distinguish between bleedings at different positions adjacent to the skull bone for a similar but simplified problem. The classification accuracy is 94-100% for all classes, a result that encourages us to pursue our efforts with MWT for more realistic scenarios. This indicates that MWT has potential for localizing a detected bleeding, which would increase the diagnostic value of this technique.

Cerebrovascular mechanical properties and slow waves of intracranial pressure in TBI patients.

Myogenic autoregulation of cerebral blood flow is one of the mechanisms affecting cerebral hemodynamics. Short or long-lasting changes in intracranial pressure (ICP) are believed to reveal the responses of the cerebral system to myogenic stimuli. Through the incorporation of a theoretical model into the experimental measurements of cerebrovascular distensibility and compliance in patients with traumatic brain injury (TBI), the current study is an attempt to explain ICP dynamics in either presence or absence of cerebral autoregulation. The pulse wave velocity and transfer function between arterial blood pressure and ICP were utilized as the major tools to reflect variations in the mechanical properties of distant cerebral artries/arteriols. The results imply that different states of cerebral autoregulation and associated regimes within the cerebrovascular system can lead to different types of interrelationship between the slow variations of ICP, cerebral arterial distensibility, and compliance. Consequently, each of these classes may require different types of treatment on patients with TBI.

A multidimensional signal processing approach for classification of microwave measurements with application to stroke type diagnosis.

A multidimensional signal processing method is described for detection of bleeding stroke based on microwave measurements from an antenna array placed around the head of the patient. The method is data driven and the algorithm uses samples from a healthy control group to calculate the feature used for classification. The feature is derived using a tensor approach and the higher order singular value decomposition is a key component. A leave-one-out validation method is used to evaluate the properties of the method using clinical data.

A comparison between the transfer function of ABP to ICP and compensatory reserve index in TBI.

BACKGROUND: The transfer functions which map the arterial blood pressure to the intracranial pressure and the compensatory reserve index have been investigated by various groups to evaluate the brain compliance of patients with traumatic brain injury. The focus of this study has been to assess the capability of both the above mentioned methods to monitor the intracranial compliance in patients suffering from brain swelling. MATERIALS AND METHODS: Clinical data was collected from sixteen traumatic brain injury patients and split into 4 min segments. For each segment, both the magnitude of the empirical transfer function at the fundamental cardiac frequency and the compensatory reserve index were extracted. FINDINGS: The mean values of the compensatory reserve index and the magnitude of the transfer function which scored higher than 0.7 and 0.1 respectively were recorded for all patients suffering from brain swelling. By comparing the histogram of the magnitude of the transfer function at the fundamental cardiac frequency with the histogram of the compensatory reserve index for all patients, a positive correlation between the mean values and a negative correlation among their variances were observed. The linear correlation between the mean values was estimated at r = 0.82 (p < 0.0001). CONCLUSIONS: These observations suggest that to evaluate the intracranial compensatory reserve, the magnitude of 0.1 could be a useful threshold for the transfer function at the fundamental cardiac frequency.

Harmonics tracking of intracranial and arterial blood pressure waves.

Considering cardiorespiratory interaction and heart rate variability, a new approach is proposed to decompose intracranial pressure and arterial blood pressure to their different harmonics. The method is based on tracking the amplitudes of the harmonics by a Kalman filter based tracking algorithm. The algorithm takes benefit of combined frequency estimation technique which uses both Fast Fourier Transform and RR-interval detection. The result would be of use in intracranial pressure and arterial blood pressure waveform analysis as well as other investigations which need to estimate contribution of specific harmonic in above mentioned signals such as Pressure-Volume Compensatory Reserve assessment.

Frequency interpretation of tidal peak in intracranial pressure wave.

A new approach to locate different components of ICP signal for each cardiac induced ICP beat is presented. In this method an initial timing map is used to define the appropriate part of the ICP wave which should be searched for the specific component. In parallel a recently proposed method was used to decompose the ICP wave to its different frequency harmonics. This algorithm, which is based on tracking the amplitude of the harmonic components using Kalman filtering, brings both heart rate variability and cardiorespiratory interaction into account and provides good time and frequency resolution. Comparing the results of two methods for seventeen ICP records, each one hour long, it has been observed that the fundamental cardiac component has the most significant contribution in the construction of the tidal peak in ICP and therefore tracking of this harmonic could be informative of the tidal peak evolution over the time.