Photic driving response in primary headache: diagnostic value tested by discriminant analysis and artificial neural network classifiers

Systematic Review of the Effectiveness of Machine Learning Algorithms for Classifying Pain Intensity, Phenotype or Treatment Outcomes Using Electroencephalogram Data

Recent attempts to utilize machine learning (ML) to predict pain-related outcomes from Electroencephalogram (EEG) data demonstrate promising results. The primary aim of this review was to evaluate the effectiveness of ML algorithms for predicting pain intensity, phenotypes or treatment response from EEG. Electronic databases MEDLINE, EMBASE, Web of Science, PsycINFO and The Cochrane Library were searched. A total of 44 eligible studies were identified, with 22 presenting attempts to predict pain intensity, 15 investigating the prediction of pain phenotypes and seven assessing the prediction of treatment response. A meta-analysis was not considered appropriate for this review due to heterogenos methods and reporting. Consequently, data were narratively synthesized. The results demonstrate that the best performing model of the individual studies allows for the prediction of pain intensity, phenotypes and treatment response with accuracies ranging between 62 to 100%, 57 to 99% and 65 to 95.24%, respectively. The results suggest that ML has the potential to effectively predict pain outcomes, which may eventually be used to assist clinical care. However, inadequate reporting and potential bias reduce confidence in the results. Future research should improve reporting standards and externally validate models to decrease bias, which would increase the feasibility of clinical translation. PERSPECTIVE: This systematic review explores the state-of-the-art machine learning methods for predicting pain intensity, phenotype or treatmentresponse from EEG data. Results suggest that machine learning may demonstrate clinical utility, pending further research and development. Areas for improvement, including standardized processing, reporting and the need for better methodological assessment tools, are discussed.

Brain Excitability in Tension-Type Headache: a Separate Entity from Migraine?

PURPOSE OF REVIEW

Tension-type headache is often regarded as the "normal" headache due to its high prevalence and mild disability in contrast with migraine. Clinically, both headaches are common comorbidities to each other. To date there has been many studies linked migraine to a brain excitability disorder. This review summarized earlier studies on brain excitability of TTH and discuss if TTH is a separate clinical entity from migraine as suggested by the diagnostic criteria.

RECENT FINDINGS

A recent magnetoencephalographic study from our group enrolled patients with "strict-criteria" TTH (i.e., absence of any migraine characteristics and associated symptoms) to compare the somatosensory excitability with patients with migraine and controls. This study provided evidence that TTH and migraine differ in excitability profiles and the measurement of preactivation excitability was able to discriminate TTH from migraine. Earlier studies on brain excitability of TTH yielded negative findings or a common change shared with migraine. Future studies using strict diagnostic criteria to avoid the unwanted interference from migraine comorbidity may help decipher the "true" pathophysiology of TTH, which may pave the way to a TTH-specific brain signature and treatment.

Laboratory tests of headache disorders – dawn of a new era?

Context

The classification of headache disorders has improved over the years, but further work is needed to develop and improve headache diagnosis within headache subtypes. The present review is a call for action to implement laboratory tests in the classification and management of primary and some secondary headaches.

Background

In this narrative review we present and discuss published tests that might be useful in phenotyping and/or diagnosis of long-lasting headache disorders such as migraine, tension-type headache, trigeminal autonomic cephalalgias, trigeminal neuralgia and persisting secondary headaches.

Aim

The palpometer test, quantitative sensory testing, nociceptive blink reflex and autonomic tests may be valuable to phenotype and/or diagnose subforms of migraine, tension-type headache, cluster headache, trigeminal neuralgia and medication-overuse headache. Provocation tests with glyceryl trinitrate (GTN) and calcitonin gene-related peptide (CGRP) may be valuable in subclassification of migraine and cluster headache. Lumbar pressure monitoring and optical coherence tomography may valuable tools to diagnose and follow patients with chronic headache and raised intracranial pressure.

Finding

A number of laboratory tests in headache research are presently available, but have primarily been performed in single research studies or a few studies that differ in methods and patient groups. At present, there is no evidence-based strategy for implementing diagnostic tests, but this could be achieved if well-reputed tertiary headache centers commence developing and implementing laboratory tests in order to improve the classification and treatment of headache patients.

Analysis and clinical correlates of 20 Hz photic driving on routine EEG in migraine

Enhanced photic driving (PD) during high-frequency flicker stimulation, the so-called H response, is a classical feature of migraine patients between attacks, but is thought to be of poor clinical utility. Visual inspection of the EEG for its detection may not be reliable, however, data on its possible correlations with clinical features and migraine pathophysiology are scarce. We have compared visual inspection and EEG spectral analysis to detect abnormal PD in 280 consecutive migraine patients of our headache clinic (episodic migraine without aura, n = 171; chronic migraine, n = 48; migraine with aura, n = 61) and in a group of 24 non-migrainous neurological controls. Spectral frequency analyses were performed blindly by one of us (YF). On visual inspection, 50.4 % of migraineurs were thought to have increased 20 Hz PD. After spectral analysis, only 62.4 % of them had PD power superior to the mean + 95 % CI of the control group. Sensitivity of visually identified PD was 82.24 %, specificity 69.36 %. Increased PD on spectral analysis was more prevalent in episodic migraine than in chronic migraine, in patients with low attack frequency, in those with ictal autonomic symptoms in addition to nausea and in those with a strong family history of migraine. We confirm therefore that 20 Hz photic driving is of little diagnostic utility and its prevalence in migraine overestimated on visual inspection. Its presence on spectral analysis of the EEG, however, might be of pathophysiological interest, as it identifies subgroups of migraineurs of whom the common denominator could be lack of habituation of cortical responses during repetitive stimulation.

Chirp stimulation: H-response short and dynamic

Background

Migraine patients have an increased photic-driving response. This ‘H-response’ (HR) has potential diagnostic value but it is time consuming.

Aim

The aim of the study was to establish a fast and standardized test for the study of migraine biology and treatment.

Methods

We studied 11 migraine patients and 11 matched control participants. We used stroboscope flashes with a ‘chirp’-like linear frequency-increase from 10 to 40 Hz. EEG was recorded from occipital electrodes. Power spectral density was calculated for the stimulus response and corrected for baseline. An HR-estimator was calculated as the average power between 18 and 26 Hz in the stimulation-frequency window.

Results

There was a significant difference for single ( p < 0.05) and for 10 averaged recordings ( p < 0.01) between migraineurs and controls, and a high inter-test reliability (Cronbach’s alpha = 0.94).

Conclusion

Chirp-like stimulation to study the H-response is reliable and efficient and might therefore have a potential for acute interventional studies in migraine research.

Characteristic electroencephalographic findings by photic driving in patients with migraine-associated vertigo

CONCLUSIONS

Patients with migraine-associated vertigo (MAV) may have a higher incidence of photic-driven electroencephalography (EEG) response. The occurrence of photic-driven responses for stimulation at frequencies of 10, 12, and 15 Hz can be considered a positive indicator for MAV. Although photic driving is not specific to migraine, patients with MAV have a higher incidence of photic driving. The distinction between Ménière's disease and MAV by means of EEG would be another interesting topic of research.

OBJECTIVE

The objective of this study was to distinguish between patients with MAV and patients with other vestibulopathies (VPs) using EEG.

METHODS

EEG with intermittent photic stimulation (PS) at frequencies between 3 and 24 Hz was recorded to quantify photic driving in patients with MAV (n = 21) and VPs (n = 15). The presence or absence of photic driving was determined; a score of 1 was counted for response to a particular frequency.

RESULTS

Total scores were calculated for each patient for statistical comparison. There was a significant difference in the average total score between MAV and VP patients (p < 0.05), and photic-driven responses were observed for 10, 12, and 15 Hz stimulation.

Resilience in migraine brains: decrease of coherence after photic stimulation

Background: During migraine attacks, patients generally have photophobia and phonophobia and seek for environments with less sensorial stimulation. Present work aimed to quantify cortical partial directed coherence (PDC) of electroencephalographic (EEG) recordings from migraine patients and controls in occipital, parietal, and frontal areas with or without photic stimulation. Our hypothesis is that migraine patients with visual aura might have neuronal networks with higher coherence than controls even in interictal periods due to a predisposition in sensory cortical processing. Methods: Eleven adult women with migraine with visual aura (at least 48 h without previous attacks) and seven healthy adult woman were submitted to EEG recording in basal state and during photic stimulation. Results: When compared to healthy volunteers, migraine patients show different coherence profiles. Migraine patients had greater coherence than controls during the basal period (without photic stimulation), showing predisposition for sensory processing in many frequency ranges. After photic stimulation, patients showed a decrease in cortical coherence while controls had an increase. Conclusions: When compared to healty subjects, migraineurs show increased cortical coherence before photic stimulation, but a decrease when stimulation starts. This may be the expression of a resilience mechanism that allows migraineurs the interictal period. The PDC analysis permits to address a patient coherence profile, or “coherence map,” that can be utilized for management of the headache disorder or following up treatments.

Effects of levetiracetam vs topiramate and placebo on visually evoked phase synchronization changes of alpha rhythm in migraine

OBJECTIVE

Recent theories about migraine pathogenesis have outlined an abnormal central processing of sensory signals, also suggested by an abnormal pattern of EEG hyper-synchronization under visual stimulation. The aim of the present study was to test the efficacy of topiramate and levetiracetam vs placebo in a double blind project observing the effects of the three treatments on the EEG synchronization in the alpha band under sustained flash stimulation.

METHODS

Forty-five migraine without aura outpatients (MO) were selected and randomly assigned to 100mg topiramate, 1000 mg levetiracetam or placebo treatment. In addition, 24 non-migraine healthy controls were submitted to EEG analysis. The EEG was recorded by 19 channels: flash stimuli with a luminosity of 0.2J were delivered, in a frequency range from 3 to 30 Hz. We evaluated the phase synchronization index, that we previously applied in migraine, after EEG signals filtering in the alpha band. Our approach was based on the Hilbert transform.

RESULTS

Both levetiracetam and topiramate significantly decreased migraine frequency, compared with placebo. MO patients displayed increased alpha-band phase synchronization as an effect of stimulus frequency; on the other hand the stimuli had an overall desynchronizing effect on control subjects. The phase synchronization index separates the two stages, before and after the treatment, only for levetiracetam, at stimulus frequencies of 9, 18, 24 and 27 Hz.

CONCLUSIONS

An abnormal alpha band synchronization under visual stimuli was confirmed in migraine; this phenomenon was reversed by levetiracetam preventive treatment.

SIGNIFICANCE

These results confirmed in humans the inhibiting action of levetiracetam on neuronal hyper-synchronization.

Physiological Parameters as Biomarkers of Migraine

We review physical signs and measurements from neurophysiological research, which may be considered biomarkers of migraine. Most studies show that information processing is abnormal in migraineurs. Studies of trigeminal nociception are in line with a central sensitization during the attack. The best documented abnormality is an interictal lack of habituation in migraineurs, with a normalization during the headache state, which has been shown using evoked and event-related potentials. Together with a diminished energy reserve found in MR-spectroscopy studies, these results suggest a possible role of increased energy consumption in attack generation. Importantly, to date, no neurophysiological marker has a high enough sensitivity and specificity to point out the single migraine patient. Therefore, these biomarkers are to be understood as tools for research and are not to be used for diagnostic purposes.

Prediction of survival in surgical unresectable lung cancer by artificial neural networks including genetic polymorphisms and clinical parameters

Lung cancer, a common malignancy in Taiwan, involves multiple factors, including genetics and environmental factors. The survival time is very short once cancer is diagnosed as being in advanced stage and surgically unresectable. Therefore, a good model of prediction of disease outcome is important for a treatment plan. We investigated the survival time in advanced lung cancer by using computer science from the genetic polymorphism of the p21 and p53 genes in conjunction with patients' general data. We studied 75 advanced and surgical unresectable lung cancer patients. The prediction of survival time was made by comparing real data obtained from follow-up periods with data generated by an artificial neural network (ANN). The most important input variable was the clinical staging of lung cancer patients. The second and third most important variables were pathological type and responsiveness to treatment, respectively. There were 25 neurons in the input layer, four neurons in the hidden layer-1, and one neuron in the output layer. The predicted accuracy was 86.2%. The average survival time was 12.44 +/- 7.95 months according to real data and 13.16 +/- 1.77 months based on the ANN results. ANN provides good prediction results when clinical parameters and genetic polymorphisms are considered in the model. It is possible to use computer science to integrate the genetic polymorphisms and clinical parameters in the prediction of disease outcome. Data mining provides a promising approach to the study of genetic markers for advanced lung cancer.

Steady-state visual evoked potentials in the low frequency range in migraine: a study of habituation and variability phenomena

Previous studies have revealed that migraine patients display an increased photic driving to flash stimuli in the medium frequency range. The aim of this study was to perform a topographic analysis of steady-state visual evoked potentials (SVEPs) in the low frequency range (3-9 Hz), evaluating the temporal behaviour of the F1 amplitude by investigating habituation and variability phenomena. The main component of SVEPs, the F1, demonstrated an increased amplitude in several channels at 3 Hz. Behaviour of F1 amplitude was rather variable over time, and the wavelet-transform standard deviation was increased in migraine patients at a low stimulus rate. The discriminative value of the F1 mean amplitude and variability index, tested by both an artificial neural network classifier and a support vector machine, were high according to both methods. The increased photic driving in migraine should be subtended by a more generic abnormality of visual reactivity instead of a selective impairment of a visual subsystem. Temporal behaviour of SVEPs is not influenced by a clear tendency to habituation, but the F1 amplitude seemed to change in a complex way, which is better described by variability phenomena. An increased variability in response to flicker stimuli in migraine patients could be interpreted as an overactive regulation mechanism, prone to instability and consequently to headache attacks, whether spontaneous or triggered.

Electroencephalogram processing using neural networks

The electroencephalogram (EEG), a highly complex signal, is one of the most common sources of information used to study brain function and neurological disorders. More than 100 current neural network applications dedicated to EEG processing are presented. Works are categorized according to their objective (sleep analysis, monitoring anesthesia depth, brain-computer interface, EEG artifact detection, EEG source-based localization, etc.). Each application involves a specific approach (long-term analysis or short-term EEG segment analysis, real-time or time delayed processing, single or multiple EEG-channel analysis, etc.), for which neural networks were generally successful. The promising performances observed are demonstrative of the efficiency and efficacy of systems developed. This review can aid researchers, clinicians and implementors to understand up-to-date interest in neural network tools for EEG processing. The extended bibliography provides a database to assist in possible new concepts and idea development.

Photic driving in the electroencephalogram of children and adolescents: harmonic structure and relation to the resting state

In order to identify latent bioelectrical oscillators, 15 normal subjects (aged 9-17 years, 8 males, 7 females) were subjected to intermittent photic stimulation. The EEG amplitude spectra corresponding to the 11 fixed frequencies of stimulation presented (3-24 Hz) were combined to form "profiles" of the driving reaction in the right occipital area. The driving response varied with frequency, and was demonstrable in 70-100% of cases (using as criterion peak amplitudes 20% larger than those of the neighbors). The strongest responses were observed at the frequency closest to the alpha peak of the resting EEG. A secondary profile maximum was in the theta band. In 10 subjects, this maximum exceeded half the alpha peak (with an average of 72.4% of the alpha peak), while in the resting spectra, theta amplitudes were much lower than the alpha maxima. This responsiveness in theta activity seems to be characteristic of prepubertal and pubertal subjects. The profiles and resting EEG spectra showed a highly significant Pearson's correlation, with the peak in the theta band of the profiles being the main difference observed between them. The correlation coefficient was significantly correlated with the ratio of the maxima in the theta and alpha bands (R = -0.77, P<0.001). The correlation coefficient between profile and resting spectrum may be a useful indicator in screening methods used to reveal latent cerebral oscillators. Profiles for the second and third harmonics were correlated with those of the first harmonic (fundamental frequency), when considering the corresponding EEG frequencies. Peak frequencies in all three profiles were close to those of the individual's background alpha rhythm, and peak amplitudes in higher harmonics were not much lower than those of the fundamental frequency (mean values of 84 and 63%, for second and third harmonics, respectively).