Spectral F-Test Research Articles

Considering the effect of correlation between the channels in multivariate detectors for evoked responses in the electroencephalogram

Evoked potentials refer to the small responses in the electroencephalogram (EEG), resulting from repeated or continuous auditory, visual or other stimuli. The detection of these responses is most commonly based on visual inspection by experienced professionals. To overcome the inherent subjectivity, automated (usually statistical) methods can be used, which aim to find the statistical significance of some parameter(s) calculated from the recorded signals, based on the null-hypothesis of 'no stimulus response is present'. These tests need to rely on assumptions about the statistical distributions of the signals and their parameters which in many cases allows p-values to be calculated. For some convenient parameters, however, closed form mathematical solutions are intractable and simulations are often required to find p-values (or critical values). The use of multichannel recordings can increase the sensitivity of tests compared to single-channel EEG, permitting the detection of responses in shorter recordings. This can be achieved by simple extensions of originally single channel methods, such as magnitude-squared coherence, component synchrony measure or the spectral F-test. However, the correlation between EEG channels can lead to false-positive rates that are difficult to predict or control. In this paper we present a modified parametric bootstrap method that permits the accurate estimation of false positive rates (p-values and critical values) for multichannel extensions of univariate objective evoked response techniques, with concomitant increases in sensitivity over conventional single channel methods. Illustrative results with evoked potentials from repeated photic stimulation are provided, with multivariate magnitude-squared coherence demonstrating the most robust performance.

Choosing multichannel objective response detectors for multichannel auditory steady-state responses

The assessment of evoked responses still largely relies on subjective assessment, though a number of objective response detection techniques have been proposed, which are usually based on the statistical analysis of electroencephalogram (EEG) signals. In the current work, we expand on this with an investigation of multichannel objective response detection (MORD) methods, by assessing the performance of six previously published MORD algorithms together with three new methods (named product of Local Spectral F-test, average and the product of Component Synchrony Measure). The MORD techniques were evaluated in both simulated data and in multifrequency Auditory Stead-State Responses (ASSRs) recorded from 11 normal-hearing volunteers. The performances of MORD techniques were evaluated based on detection rates and the number of false positive tests. The results demonstrate that inflated false positive rates can arise when correlation between EEG channels is not taken into account by some of the MORD methods, but the algorithms are fairly robust to low cross-channel correlations. Furthermore, two of the already established MORD techniques showed immunity to cross-channel correlation. The results showed that the sensitivity of response detection increased when more than one channel was used, and an increase of up to 16 % compared to the best single-channel method could be achieved by the newly proposed best MORD methods. MORD methods are thus a promising new tool for ASSR detection.

Assessment of high-frequency steady-state visual evoked potentials from below-the-hairline areas for a brain-computer interface based on Depth-of-Field

Background and ObjectiveRecently, a promising Brain-Computer Interface based on Steady-State Visual Evoked Potential (SSVEP-BCI) was proposed, which composed of two stimuli presented together in the center of the subject’s field of view, but at different depth planes (Depth-of-Field setup). Thus, users were easily able to select one of them by shifting their eye focus. However, in that work, EEG signals were collected through electrodes placed on occipital and parietal regions (hair-covered areas), which demanded a long preparation time. Also, that work used low-frequency stimuli, which can produce visual fatigue and increase the risk of photosensitive epileptic seizures. In order to improve the practicality and visual comfort, this work proposes a BCI based on Depth-of-Field using the high-frequency SSVEP response measured from below-the-hairline areas (behind-the-ears). MethodsTwo high-frequency stimuli (31 Hz and 32 Hz) were used in a Depth-of-Field setup to study the SSVEP response from behind-the-ears (TP9 and TP10). Multivariate Spectral F-test (MSFT) method was used to verify the elicited response. Afterwards, a BCI was proposed to command a mobile robot in a virtual reality environment. The commands were recognized through Temporally Local Multivariate Synchronization Index (TMSI) method. ResultsThe data analysis reveal that the focused stimuli elicit distinguishable SSVEP response when measured from hairless areas, in spite of the fact that the non-focused stimulus is also present in the field of view. Also, our BCI shows a satisfactory result, reaching average accuracy of 91.6% and Information Transfer Rate (ITR) of 5.3 bits/min. ConclusionThese findings contribute to the development of more safe and practical BCI.

A bayesian approach to the spectral F-Test: Application to auditory steady-state responses

Background and ObjectiveAuditory steady-state responses (ASSRs) represent an objective method used in clinical practice to assess hearing thresholds. The steady-state nature of these signals allows response detection by means of statistical techniques in the frequency domain as spectral F-test. This objective response detection (ORD) compares the power of the response bin against the power of the neighboring frequency noise bins. Most ORD algorithms are based on the Neyman-Pearson approach to the hypothesis test provided that the likelihood ratio test is the most powerful test for a given significance level alpha (also called Type I error). On the other hand, the Bayesian approach allows the inclusion of prior information in the model and enables the updating of this information with posterior knowledge. This approach, however, has not been explored with respect to ORD techniques, thus enabling the exploration of new paradigms, which may contribute to this field of study, especially in terms of the time required for response detection. The aim of this study is to use the Bayesian approach in the implementation of the spectral F-test for application to ASSRs. MethodsMonte Carlo simulations were performed to evaluate Neyman-Pearson and Bayesian detectors’ performances with the spectral F-test as a function of the signal-to-noise ratio. Then, the two detectors were applied to ASSR recordings of nine normal-hearing individuals subjected to amplitude-modulated tones of various intensities. ResultsBoth simulation and ASSR data analyses showed that among the scenarios analyzed, the most promising case was that in which the lowest possible values for the a priori probability were selected for the null hypothesis (H0), allowing detection at low signal-to-noise ratios. The worst performance occurred when the a priori probabilities for both hypotheses were equal. The ASSR data also showed that higher stimulus intensity led to better performance and faster detection due to improvements in the signal-to-noise ratio. ConclusionsThe a priori probabilities can affect the Bayesian detector's performance, directly impacting the time needed to identify responses. The parallel behaviors observed between the performances of both approaches showed that the Bayesian detector can achieve its ideal performance at lower signal-to-noise ratios compared to the optimal performance of the Neyman-Pearson detector, reflecting the promising applicability of the Bayesian approach to evoked potentials.

Multivariate approach for estimating the local spectral F-test and its application to the EEG during photic stimulation

Background and objectiveThe local spectral F-test (SFT) corresponds to a statistical way of assessing whether the spectrum of a signal is flat in the vicinity of a specific frequency. The power of this univariate test (comparing one frequency component against its neighbours using only one signal) depends on the signal-to-noise ratio, which is fixed in the case of electroencephalogram (EEG) analysis. However, this limitation could be overcome by considering more signals in the analysis. Thus, this work presents an alternative multivariate approach for estimating the local SFT. MethodsProbabilities of detection and false alarm studies were performed for this new detector using Monte Carlo simulations and theoretically whenever possible. The application was illustrated in recorded EEG data collected during photic stimulation. ResultsThe results showed that it is worth using more channels if available, since the probability of detecting a response tends to increase with increasing number of signals. In the application to the EEG during photic stimulation, the best results were obtained by using N > 2 signals (around 30% more accurate when compared with the univariate case. The false positive levels were maintained below 5%). ConclusionConsequently, it is conjectured that it is always better to apply the proposed method if more than one EEG signal with the same signal-to-noise ratio (SNR) is available. For the case where the SNRs are different, a guideline has been given to improve the detection.

Improving the detection of evoked responses to periodic stimulation by using bivariate local spectral F-test – Application to EEG during photic stimulation

The spectral local F-test has been applied for detecting evoked responses to rhythmic stimulation that are embedded in the ongoing electroencephalogram (EEG). Based on the sampling distribution of a flat spectrum at the neighbourhood of the stimulation frequency, spectral peaks in an EEG signal that are due to the stimulation may be readily assessed. Nevertheless, the performance of the technique is strongly affected by both the signal-to-noise ratio (SNR) of the responses and the number of data segments used in the estimation. The present work aims at both deriving and evaluating a multivariate extension of local F-test by including the EEG collected at a second distinct derivation. The detection rate with this multivariate detector was found to be greater than that using a single channel in case of equal SNR in both signals. Monte Carlo simulation results showed that the probability of detection with this new detector saturates for signal-to-noise ratios above 12 dB and indicated a greater detection rate in practical situations, even when smaller SNR-values are found in the added signal (e.g. 5 dB for 16 neighbouring frequencies used in the estimation). The technique was next applied to the EEG from 12 subjects during intermittent, photic stimulation leading to superior performance in comparison with the univariate local F-test. Since a higher detection rate with the proposed technique is achieved without the need of increasing the number of data segments, it allows evoked responses to be detected faster, once the same detection rate may be accomplished with less segments. This might be useful in clinical practice.

Assessment of frequency specific auditory steady-state response using amplitude modulation with 2-order exponential envelope.

This study investigated the performance of Frequency Specific Auditory Steady-State Response (FS-ASSR) detection elicited by the amplitude modulated tone with 2-order exponential envelope (AM2), using objective response detection (ORD) techniques of Spectral F-Test (SFT) and Magnitude Squared Coherence (MSC). ASSRs from 24 normal hearing adults were obtained during binaural multi-tone stimulation of amplitude-modulation (AM) and AM2 at intensities of 60, 45 and 30 dBSPL. The carrier frequencies were 500, 1000, 2000, and 4000 Hz, modulated between 77 and 105 Hz. AM2 achieve FS-ASSR amplitudes higher than AM by 16%, 18% and 12% at 60, 45 and 30 dBSPL, respectively, with a major increase at 500 Hz (22.5%). AMS2PL increased the Detection Rate (DR) up to 8.3% at 500 Hz for 30 dBSPL, which is particularly beneficial for FS-ASSR detection near the hearing threshold. In addition, responses in 1000 and 4000 Hz were consistently increased. The MSC and SFT presented no differences in Detection Rate (DR). False Detection Rate (FDR) was close to 5% for both techniques and tones. Detection times to reach DR over 90% were 3.5 and 4.9 min at 60 and 45 dBSPL, respectively. Further investigation concerning efficient multiple FS-ASSR is still necessary, such as testing subjects with hearing loss.

Detection efficiency of auditory steady state evoked by modulated noise

AimThis study aimed to investigate the efficiency of Magnitude Squared Coherence (MSC) and Spectral F test (SFT) for the detection of auditory steady state responses (ASSR) obtained by amplitude-modulated noises. Material and methodsTwenty individuals (12 women) without any history of neurological or audiological diseases, aged from 18 to 59 years (mean ± standard deviation = 26.45 ± 3.9 years), who provided written informed consent, participated in the study. The Audiostim system was used for stimulating and ASSR recording. The tested stimuli were amplitude-modulated Wide-band noise (WBN), Low-band noise (LBN), High-band noise (HBN), Two-band noise (TBN) between 77 and 110 Hz, applied in intensity levels of 55, 45, and 25 dB sound pressure level (SPL). MSC and SFT, two statistical-based detection techniques, were applied with a significance level of 5%. Detection times and rates were compared using the Friedman test and Tukey-Kramer as post hoc analysis. Also based on the stimulation parameters (stimuli types and intensity levels) and detection techniques (MSC or SFT), 16 different pass/fail protocols, for which the true negatives (TN) were calculated. ResultsThe median detection times ranged from 68 to 157s for 55 dB SPL, 68–99s for 45 dB SPL, and 84–118s for 25 dB SPL. No statistical difference was found between MSC and STF considering the median detection times (p > 0.05). The detection rates ranged from 100% to 55.6% in 55 dB SPL, 97.2%–38.9% in 45 dB SPL and 66.7%–8.3% in 25 dB SPL. Also for detection rates, no statistical difference was observed between MSC and STF (p > 0.05). True negatives (TN) above 90% were found for Protocols that employed WBN or HBN, at 55 dB SPL or that used WBN or HBN, at 45 dB SPL. For Protocols employing TBN, at 55 dB SPL or 45 dB SPL TN below 60% were found due to the low detection rates of stimuli that included low-band frequencies. ConclusionThe stimuli that include high-frequency content showed higher detection rates (>90%) and lower detection times (<3 min). The noise composed by two bands applied separately (TBN) is not feasible for clinical applications since it requires prolonging the exam duration, and also led to a reduced percentage of true negatives.On the other hand, WBN and HBN achieved high detection performance and high TN and should be investigated to implement pass/fail protocol for hearing screening with clinical population. Finally, both WBN and HBN seemed to be indifferent to the employed technique (SFT or MSC), which can be seen as another advantage of ASSR employment.

Multivariate evoked response detection based on the spectral F-test

BackgroundObjective response detection techniques, such as magnitude square coherence, component synchrony measure, and the spectral F-test, have been used to automate the detection of evoked responses. The performance of these detectors depends on both the signal-to-noise ratio (SNR) and the length of the electroencephalogram (EEG) signal. New methodRecently, multivariate detectors were developed to increase the detection rate even in the case of a low signal-to-noise ratio or of short data records originated from EEG signals. In this context, an extension to the multivariate case of the spectral F-test detector is proposed. ResultsThe performance of this technique is assessed using Monte Carlo. As an example, EEG data from 12 subjects during photic stimulation is used to demonstrate the usefulness of the proposed detector. Comparison with existing method(s)The multivariate method showed detection rates consistently higher than those ones when only one signal was used. ConclusionsIt is shown that the response detection in EEG signals with the multivariate technique was statistically significant if two or more EEG derivations were used.

Statistical evaluation of a novel SSVEP-BCI stimulation setup based on depth-of-field

Introduction The main drawback of a Brain-computer Interface based on Steady-State Visual Evoked Potential (SSVEP-BCI) that detects the emergence of visual evoked potentials (VEP) in reaction to flickering stimuli is its muscular dependence due to users must redirect their gaze to put the target stimulus in their field of view. In this work, a novel setup is evaluated in which two stimuli are placed together in the center of users' field of view, but with dissimilar distances from them, so that the target selection is performed by focus shifting instead of head, neck and/or eyeball movements. Methods A model of VEP generation for the novel setup was developed. The Spectral F-test based on Bartett periodogram was used to evaluate the null hypothesis of absence of effects of the non-focused stimulus (NFS) within the VEP elicited by the focused stimulus (FS). To reinforce that there is not statistical evidence to support the presence of NFS effects, the PSDA detection method was employed to find the frequency of FS. Electroencephalographic signals of nine subjects were recorded. Results Approximately in 80% of the tests, the null hypothesis with 5% level of significance was non-rejected at the fundamental frequency of NFS. The average of the accuracy rate attained with PSDA detection method was 79.4%. Conclusion Results of this work become further evident to state that if the focused stimulus (FS) will be able to elicit distinguishable VEP pattern regardless the non-focused stimulus (NFS) is also present.

Identification of drowsiness and alertness conditions by means of Spectral F-Test applied to pupillometric signals

The autonomous regulation of pupil size provides an objective physiological measure of alertness level. Pupil diameter decreases and its fluctuations increase during drowsiness. In this work, a statistical based technique in frequency domain, known as Spectral F-Test (SFT), was applied in order to compare the power of pupillometric signals measured in two conditions, wakefulness and sleepiness, from eleven volunteers. SFT was calculated based on Welch Periodograms of time series of pupil diameter (TSPD) signals. The median power for TSPD in two conditions and each volunteer at frequencies bellow 0.8 Hz were compared by means of the paired Wilcoxon signed-rank test. The percentages of volunteers for whom power during drowsiness were statistically higher than power during alertness achieved 100% for several frequencies bellow 0.2 Hz. For frequencies from 0.2 to 0.8 Hz, the percentages presented high variability, fluctuating from 40 to 90%. The Wilcoxon test indicated that median power of TSPD during drowsiness was higher than TSPD during alertness for frequencies bellow 0.2 Hz presented statistical difference. Hence, the SFT showed to be a suitable tool for drowsiness/alertness differentiation that could be applied to sleep studies.

Spectral F-Test power evaluation in the EEG during intermittent photic stimulaton

Intermittent photic stimulation (IPS) is an important functional test, which can induce the photic driving in the electroencephalogram (EEG). It is capable of enhancing latent oscillations manifestations not present in the resting EEG. However, for adequate quantitative evaluation of the photic driving, these changes should be assessed on a statistical basis. With this aim, the sampling distribution of spectral F test was investigated. On this basis, confidence limits of the SFT-estimate could be obtained for different practical situations, in which the signal-to-noise ratio and the number of epochs used in the estimation may vary. The technique was applied to the EEG of 10 normal subjects during IPS, and allowed detecting responses not only at the fundamental IPS frequency but also at higher harmonics. It also permitted to assess the strength of the photic driving responses and to compare them in different derivations and in different subjects.

Topographic aspects of photic driving in the electroencephalogram of children and adolescents.

The electroencephalogram amplitude spectra at 11 fixed frequencies of intermittent photic stimulation of 3 to 24 Hz were combined into driving "profiles" for 14 scalp points in 8 male and 7 female normal subjects aged 9 to 17 years. The driving response varied over frequency and was detected in 70 to 100% of cases in the occipital areas (maximum) and in 27 to 77% of cases in the frontal areas (minimum) using as a criterion peak amplitude 20% higher than those of the neighbors. Each subject responded, on average, to 9.7 +/- 1.15 intermittent photic stimulation frequencies in the right occipital area and to 6.8 +/- 1.97 frequencies in the right frontal area. Most of the driving responses (in relation to the previous background) were significant according to the spectral F-test (alpha = 0.05), which also detected changes in some cases of low amplitude responses not revealed by the peak criterion. The profiles had two maxima in the alpha and theta bands in all leads. The latter was not present in the background spectra in the posterior areas and was less pronounced in the anterior ones. The weight of the profile theta maximum increased towards the frontal areas where the two maxima were similar, while the profile amplitudes decreased. The profiles repeated the shape of the background spectra, except for the theta band. The interhemispheric correlation between profiles was high. The theta driving detected in all areas recorded suggests a generalized influence of the theta generators in prepubertal and pubertal subjects.

Objective response detection in an electroencephalogram during somatosensory stimulation.

Techniques for objective response detection aim to identify the presence of evoked potentials based purely on statistical principles. They have been shown to be potentially more sensitive than the conventional approach of subjective evaluation by experienced clinicians and could be of great clinical use. Three such techniques to detect changes in an electroencephalogram (EEG) synchronous with the stimuli, namely, magnitude-squared coherence (MSC), the phase-synchrony measure (PSM) and the spectral F test (SFT) were applied to EEG signals of 12 normal subjects under conventional somatosensory pulse stimulation to the tibial nerve. The SFT, which uses only the power spectrum, showed the poorest performance, while the PSM, based only on the phase spectrum, gave results almost as good as those of the MSC, which uses both phase and power spectra. With the latter two techniques, stimulus responses were evident in the frequency range of 20-80 Hz in all subjects after 200 stimuli (5 Hz stimulus frequency), whereas for visual recognition at least 500 stimuli are usually applied. Based on these results and on simulations, the phase-based techniques appear promising for the automated detection and monitoring of somatosensory evoked potentials.