Fast Band Research Articles

Predictive modeling based on functional connectivity of interictal scalp EEG for infantile epileptic spasms syndrome

ObjectiveThis study aims to delineate the electrophysiological variances between patients with infantile epileptic spasms syndrome (IESS) and healthy controls and to devise a predictive model for long-term seizure outcomes. MethodsThe cohort consisted of 30 individuals in the seizure-free group, 23 in the seizure-residual group, and 20 in the control group. We conducted a comprehensive analysis of pretreatment electroencephalography, including the relative power spectrum (rPS), weighted phase-lag index (wPLI), and network metrics. Follow-up EEGs at 2 years of age were also analyzed to elucidate physiological changes among groups. ResultsInfants in the seizure-residual group exhibited increased rPS in theta and alpha bands at IESS onset compared to the other groups (all p < 0.0001). The control group showed higher rPS in fast frequency bands, indicating potentially enhanced cognitive function. The seizure-free group presented increased wPLI across all frequency bands (all p < 0.0001). Our predictive model utilizing wPLI anticipated long-term outcomes at IESS onset (area under the curve 0.75). ConclusionOur findings demonstrated an initial “hypersynchronous state” in the seizure-free group, which was ameliorated following successful treatment. SignificanceThis study provides a predictive model utilizing functional connectivity and insights into the diverse electrophysiology observed among outcome groups of IESS.

Effect of early postmenopause and premenopause on resting-state electroencephalographic and their correlation with ovarian hormone levels.

This study aimed to compare the effect of the early postmenopausal period on resting-state electroencephalographic spectral power with that of the premenopausal period and to analyze the correlation between electroencephalographic spectral power values and endogenous ovarian hormone levels. This study involved 13 early postmenopausal women and 10 premenopausal women in the early follicular, 10 in the ovulatory phase, and 10 in the early luteal phase who underwent resting-state quantitative electroencephalographic spectral power with eyes closed and endogenous ovarian hormone measurements. The delta, theta, alpha1, alpha2, beta1, and beta2 absolute power were compared between the early postmenopausal and premenopausal groups. Correlations between electroencephalographic spectral power values and 17β estradiol, progesterone, follicle-stimulating hormone (FSH), and luteinizing hormone levels were analyzed in early postmenopausal women. Compared with the premenopausal group, the early postmenopausal group showed significantly higher resting-state theta power in the frontal region, alpha1 and alpha2 power in the frontal and central regions, beta1 power in the frontal, central, parietal, and occipital regions, and beta2 power in the centroparietal region. Beta2 power values were positively correlated with FSH levels. The current findings highlight that early postmenopausal women show greater resting-state alpha and beta power, which suggests cortical excitability of fast frequency bands involved in states of alertness, focus of attention, cognition, and emotion. Additionally, we emphasized the effect of FSH levels on fast cortical activation in early postmenopausal women.

Electroencephalografic Activity during the Reading of Erotic Texts with and without Aggression

Sexual arousal (SA) is a multidimensional experience that includes cognitive, emotional, motivational and physiological components. Texts with erotic content have been used to generate a state of SA. Erotic texts often include aggressive content that has not been evaluated in relation to SA. The aim of this work was to compare cortical functionality in women when reading a sexually explicit text (SET) and a sexually explicit text with aggression (SETA). Twenty-seven women participated. The EEG activity of the frontal, temporal and parietal locations was recorded during the reading of both texts. The participants found the SET to be more pleasant than the SETA. Both texts were identified as triggers of general and SA. While reading the SETA, there was an increase in absolute power in the frontal and parietal locations, a higher intrahemispheric correlation between the left frontal and temporal locations in fast frequency bands and a greater interhemispheric correlation between the frontal locations in the delta and alpha1 bands. These findings indicate that cortical functionality during SA in women differs based on the content and context of the erotic material being read, possibly associated with mechanisms that underlie the processing and incentive value assignment of stimuli with sexual and aggressive connotations.

Sound source classification for soundscape analysis using fast third-octave bands data from an urban acoustic sensor networka).

The exploration of the soundscape relies strongly on the characterization of the sound sources in the sound environment. Novel sound source classifiers, called pre-trained audio neural networks (PANNs), are capable of predicting the presence of more than 500 diverse sound sources. Nevertheless, PANNs models use fine Mel spectro-temporal representations as input, whereas sensors of an urban noise monitoring network often record fast third-octaves data, which have significantly lower spectro-temporal resolution. In a previous study, we developed a transcoder to transform fast third-octaves into the fine Mel spectro-temporal representation used as input of PANNs. In this paper, we demonstrate that employing PANNs with fast third-octaves data, processed through this transcoder, does not strongly degrade the classifier's performance in predicting the perceived time of presence of sound sources. Through a qualitative analysis of a large-scale fast third-octave dataset, we also illustrate the potential of this tool in opening new perspectives and applications for monitoring the soundscapes of cities.

Aberrant Brain Networks and Relative Band Power in Patients with Acute Anti-NMDA Receptor Encephalitis: A Study of Resting-State EEG.

The alterations of the functional network (FN) in anti-N-methyl-Daspartate receptor (NMDAR) encephalitis have been recognized by functional magnetic resonance imaging studies. However, few studies using the electroencephalogram (EEG) have been performed to explore the possible FN changes in anti-NMDAR encephalitis. In this study, the aim was to explore any FN changes in patients with anti-NMDAR encephalitis. Twenty-nine anti-NMDAR encephalitis patients and 29 age- and gender-matched healthy controls (HC) were assessed using 19-channel EEG examination. For each participant, five 10-second epochs of resting state EEG with eyes closed were extracted. The cortical source signals of 84 Brodmann areas were calculated using the exact low resolution brain electromagnetic tomography (eLORETA) inverse solution by LORETA-KEY. Phase Lag Index (PLI) matrices were then obtained and graph and relative band power (RBP) analyses were performed. Compared with healthy controls, functional connectivity (FC) in the delta, theta, beta 1 and beta 2 bands significantly increased within the 84 cortical source signals of anti-NMDAR encephalitis patients (p < 0.05) and scalp FC in the alpha band decreased within the 19 electrodes. Additionally, the anti-NMDAR encephalitis group exhibited higher local efficiency and clustering coefficient compared to the healthy control group in the four bands. The slowing band RBP increased while the fast band RBP decreased in multiple-lobes and some of these changes in RBP were correlated with the modified Rankin Scale (mRS) and Mini-mental State Examination (MMSE) in anti-NMDAR encephalitis patients. This study further deepens the understanding of related changes in the abnormal brain network and power spectrum of anti-NMDA receptor encephalitis. The decreased scalp alpha FC may indicate brain dysfunction, while the increased source beta FC may indicate a compensatory mechanism for brain function in anti-NMDAR encephalitis patients. These findings extend understanding of how the brain FN changes from a cortical source perspective. Further studies are needed to detect correlations between altered FNs and clinical features and characterize their potential value for the management of anti-NMDAR encephalitis.

The neurophysiological effect of mild hypothermia in gyrencephalic brains submitted to ischemic stroke and spreading depolarizations.

Characterize the neurophysiological effects of mild hypothermia on stroke and spreading depolarizations (SDs) in gyrencephalic brains. Left middle cerebral arteries (MCAs) of six hypothermic and six normothermic pigs were permanently occluded (MCAo). Hypothermia began 1 h after MCAo and continued throughout the experiment. ECoG signals from both frontoparietal cortices were recorded. Five-minute ECoG epochs were collected 5 min before, at 5 min, 4, 8, 12, and 16 h after MCAo, and before, during, and after SDs. Power spectra were decomposed into fast (alpha, beta, and gamma) and slow (delta and theta) frequency bands. In the vascular insulted hemisphere under normothermia, electrodes near the ischemic core exhibited power decay across all frequency bands at 5 min and the 4th hour after MCAo. The same pattern was registered in the two furthest electrodes at the 12th and 16th hour. When mild hypothermia was applied in the vascular insulted hemispheres, the power decay was generalized and seen even in electrodes with uncompromised blood flow. During SD analysis, hypothermia maintained increased delta and beta power during the three phases of SDs in the furthest electrode from the ischemic core, followed by the second furthest and third electrode in the beta band during preSD and postSD segments. However, in hypothermic conditions, the third electrode showed lower delta, theta, and alpha power. Mild hypothermia attenuates all frequency bands in the vascularly compromised hemisphere, irrespective of the cortical location. During SD formation, it preserves power spectra more significantly in electrodes further from the ischemic core.

MTL neurons phase-lock to human hippocampal theta.

Memory formation depends on neural activity across a network of regions, including the hippocampus and broader medial temporal lobe (MTL). Interactions between these regions have been studied indirectly using functional MRI, but the bases for interregional communication at a cellular level remain poorly understood. Here, we evaluate the hypothesis that oscillatory currents in the hippocampus synchronize the firing of neurons both within and outside the hippocampus. We recorded extracellular spikes from 1854 single- and multi-units simultaneously with hippocampal local field potentials (LFPs) in 28 neurosurgical patients who completed virtual navigation experiments. A majority of hippocampal neurons phase-locked to oscillations in the slow (2-4 Hz) or fast (6-10 Hz) theta bands, with a significant subset exhibiting nested slow theta × beta frequency (13-20 Hz) phase-locking. Outside of the hippocampus, phase-locking to hippocampal oscillations occurred only at theta frequencies and primarily among neurons in the entorhinal cortex and amygdala. Moreover, extrahippocampal neurons phase-locked to hippocampal theta even when theta did not appear locally. These results indicate that spike-time synchronization with hippocampal theta is a defining feature of neuronal activity in the hippocampus and structurally connected MTL regions. Theta phase-locking could mediate flexible communication with the hippocampus to influence the content and quality of memories.

EEG complexity-based algorithm using Multiscale Fuzzy Entropy: Towards a detection of Alzheimer’s disease

Alzheimer’s Disease (AD) is a progressive neurodegenerative condition causing memory, attention, and language decline. Current AD diagnostic methods lack objectivity and non-invasiveness. While electroencephalography (EEG) holds promise for AD research, conventional EEG analysis methods have proven unsatisfactory. Non-linear dynamical approaches are considered more effective in assessing the brain’s complex nature. Starting from these considerations, this study presents an entropy-based algorithm utilizing Multiscale Fuzzy Entropy (MFE) as a promising, effective AD diagnostic method. Computed across 20 different time scales for a public dataset, MFE showed a significant discriminative power. Notably, a trend inversion was observed in the results: AD subjects displayed higher complexity values for slow frequency bands compared to healthy controls, while the opposite was found in fast frequency bands. These findings underscore the potential of MFE in effectively distinguishing AD patients from healthy individuals, marking a significant advance towards more objective and reliable AD diagnosis strategies.

EEG-based spatiotemporal dynamics of fast ripple networks and hubs in infantile epileptic spasms.

Infantile epileptic spasms (IS) are epileptic seizures that are associated with increased risk for developmental impairments, adult epilepsies, and mortality. Here, we investigated coherence-based network dynamics in scalp EEG of infants with IS to identify frequency-dependent networks associated with spasms. We hypothesized that there is a network of increased fast ripple connectivity during the electrographic onset of clinical spasms, which is distinct from controls. We retrospectively analyzed peri-ictal and interictal EEG recordings of 14 IS patients. The data was compared with 9 age-matched controls. Wavelet phase coherence (WPC) was computed between 0.2 and 400 Hz. Frequency- and time-dependent brain networks were constructed using this coherence as the strength of connection between two EEG channels, based on graph theory principles. Connectivity was evaluated through global efficiency (GE) and channel-based closeness centrality (CC), over frequency and time. GE in the fast ripple band (251-400 Hz) was significantly greater following the onset of spasms in all patients (P < 0.05). Fast ripple networks during the first 10s from spasm onset show enhanced anteroposterior gradient in connectivity (posterior > central > anterior, Kruskal-Wallis P < 0.001), with maximum CC over the centroparietal channels in 10/14 patients. Additionally, this anteroposterior gradient in CC connectivity is observed during spasms but not during the interictal awake or asleep states of infants with IS. In controls, anteroposterior gradient in fast ripple CC was noted during arousals and wakefulness but not during sleep. There was also a simultaneous decrease in GE in the 5-8 Hz range after the onset of spasms (P < 0.05), of unclear biological significance. We identified an anteroposterior gradient in the CC connectivity of fast ripple hubs during spasms. This anteroposterior gradient observed during spasms is similar to the anteroposterior gradient in the CC connectivity observed in wakefulness or arousals in controls, suggesting that this state change is related to arousal networks.

Linear and Non-linear Analyses of EEG in a Group of ASD Children During Resting State Condition

This study analyses the spontaneous electroencephalogram (EEG) brain activity of 14 children diagnosed with Autism Spectrum Disorder (ASD) compared to 18 children with normal development, aged 5–11 years. (i) Power Spectral Density (PSD), (ii) variability across trials (coefficient of variation: CV), and (iii) complexity (multiscale entropy: MSE) of the brain signal analysis were computed on the resting state EEG. PSD (0.5–45 Hz) and CV were averaged over different frequency bands (low-delta, delta, theta, alpha, low-beta, high-beta and gamma). MSE were calculated with a coarse-grained procedure on 67 time scales and divided into fine, medium and coarse scales. In addition, significant neurophysiological variables were correlated with behavioral performance data (Kaufman Brief Intelligence Test (KBIT) and Autism Spectrum Quotient (AQ)). Results show increased PSD fast frequency bands (high-beta and gamma), higher variability (CV) and lower complexity (MSE) in children with ASD when compared to typically developed children. These results suggest a more variable, less complex and, probably, less adaptive neural networks with less capacity to generate optimal responses in ASD children.

Sleep Continuity, Stability and Cyclic Organization Are Impaired in Insomniacs: A Case-Control Study.

The possibility of distinguishing insomniacs from good sleepers based on polysomnography (PSG) remains an open question. While these groups show modest differences in traditional PSG parameters, some studies suggest that finer measures may be more useful. Here we assess differences between good sleepers (GS), poor sleepers (PS) and insomniacs (IN) in classical PSG measures as well as in sleep continuity, stability and cyclic organization. PSG-monitored sleep (two nights) of 17 IN (diagnosed through a standard clinical interview; Pittsburgh Sleep Quality Index (PSQI) ≥ 5, Insomnia Severity Index (ISI) &gt; 14) was compared to that of 33 GS (PSQI &lt; 5) and 20 PS (PSQI ≥ 5, ISI ≤ 14). Compared to GS, IN were impaired in sleep macrostructure (sleep latency, sleep efficiency, WASO%) and in continuity, stability and organization, whereas PS only showed disrupted continuity and stability. Spindle parameters were comparable between IN and GS, but the former displayed enhanced power in fast frequency bands. Our findings support the hypothesis of a continuum between individuals with self-reported poor sleep and insomniacs. Further, they add to extant data on impaired sleep continuity, stability and organization in poor sleepers and elderly individuals, underlining the utility of including these measures in standard sleep assessments.

Variations in differential rotation and meridional flow within the Sun’s surface shear layer 1996–2022

We measure differential rotation and meridional flow in the Sun’s surface shear layer by tracking the motions of the magnetic network seen in magnetograms from SOHO/MDI and SDO/HMI over solar cycles 23, 24, and the start of 25 (1996–2022). We examine the axisymmetric flows derived from 15 to 24 daily measurements averaged over individual 27-day Carrington rotations. Variations in the differential rotation include the equatorial torsional oscillation—cyclonic flows centered on the active latitudes with slower flows on the poleward sides of the active latitudes and faster flows equatorward. The fast flow band starts at ∼45° latitude during the declining phase of the previous cycle and drifts equatorward, terminating at the equator at about the time of cycle minimum. Variations in the differential rotation also include a polar oscillation above 45° with faster rotation at cycle maxima and slower rotation at cycle minima. The equatorial variations were stronger in cycle 24 than in cycle 23 but the polar variations were weaker. Variations in the meridional flow include a slowing of the poleward flow in the active latitudes during cycle rise and maximum and a speeding up of the poleward flow during cycle decline and minimum. The slowing in the active latitudes was less pronounced in cycle 24 than in cycle 23. Polar countercells (equatorward flow) extend from the poles down to ∼60° latitude from time to time (1996–2000 and 2016–2022 in the south and 2001–2011 and 2017–2022 in the north). Both axisymmetric flows vary in strength with depth. The rotation rate increases inward while the meridional flow weakens inward.

Predicting collision cases at unsignalized intersections using EEG metrics and driving simulator platform

Unsignalized intersection collision has been one of the most dangerous accidents in the world. How to identify road hazards and predict the potential intersection collision ahead are challenging problems in traffic safety. This paper studies the feasibility of EEG metrics to forecast road hazards and presents an improved neural network model to predict intersection collision based on EEG metrics and driving behavior. It is demonstrated that EEG metrics show significant differences between collision and non-collision cases. It indicates that EEG metrics can serve as effective indicators to predict the collision probability. The drivers with higher relative power in fast frequency band (alpha and beta), lower relative power in slow frequency band (delta and theta) are more likely to have conflicts. The prediction using three machine learning models (Multi-layer perceptron (MLP), Logistic regression (LR) and Random forest (RF)) based on three input datasets (only EEG metrics, only driving behavior and combined EEG metrics with driving behavior) are compared. The results show that for single time point prediction, MLP model has the highest accuracy among three machine learning models. The model solely based on EEG metrics datasets has higher accuracy than driving behavior as well as combined datasets. However, for multi-time point prediction, the accuracy of MLP is only 73.9%, worse than LR and RF. We improved the MLP model by adding attention mechanism layer and using random forest model to select important features. As a consequence, the accuracy is greatly improved and reaches 88%. This study demonstrates the importance and feasibility of EEG signals to identify unsafe drivers ahead. The improved neural network model can be helpful to reduce intersection accidents and improve traffic safety.

Age-related changes in fast spindle clustering during non-rapid eye movement sleep and their relevance for memory consolidation.

Sleep plays a crucial role in memory consolidation. Recent data in rodents and young adults revealed that fast spindle band power fluctuates at a 0.02-Hz infraslow scale during non-rapid eye movement (NREM) sleep. These fluctuations result from a periodic temporal clustering of spindles and may modulate sleep maintenance and memory consolidation. With age, sleep undergoes substantial changes but age-related changes in spindle clustering have never been investigated. Polysomnography data were collected in 147 older (mean age ± SD: 69.3 ± 4.1 years) and 32 young-middle aged (34.5 ± 10.9 years) adults. Sleep-dependent memory consolidation was assessed in a subsample of 57 older adults using a visuospatial memory task. We analyzed power fluctuations in fast spindle frequency band, detected fast spindles, and quantified their clustering during the night separating encoding and retrieval. Fast spindle band power fluctuated at a 0.02-Hz infraslow scale in young-middle aged and older adults. However, the proportion of clustered fast spindles decreased non-linearly with age (p < .001). This effect was not mediated by NREM sleep fragmentation. The clustering level of fast spindles modulated their characteristics (p < .001). Finally, the mean size of spindle clusters was positively associated with memory consolidation (p = .036) and negatively with NREM sleep micro-arousal density (p = .033). These results suggest that clusters of fast spindles may constitute stable sleep periods promoting off-line processes such as memory consolidation. We emphasize the relevance of considering spindle dynamics, obviously impaired during aging, to understand the impact of age-related sleep changes on memory. Clinical Trial Information: Name: Study in Cognitively Intact Seniors Aiming to Assess the Effects of Meditation Training (Age-Well). URL: https://clinicaltrials.gov/ct2/show/NCT02977819?term=Age-Well&draw=2&rank=1. See STROBE_statement_AGEWELL.doc in supplementary material. Registration: EudraCT: 2016-002441-36; IDRCB: 2016-A01767-44; ClinicalTrials.gov Identifier: NCT02977819.

Characterization and separation of preterm and term spontaneous, induced, and cesarean EHG records

To improve the understanding of the underlying physiological processes that lead to preterm birth, and different term delivery modes, we quantitatively characterized and assessed the separability of the sets of early (23rd week) and later (31st week) recorded, preterm and term spontaneous, induced, cesarean, and induced-cesarean electrohysterogram (EHG) records using several of the most widely used non-linear features extracted from the EHG signals. Linearly modeled temporal trends of the means of the median frequencies (MFs), and of the means of the peak amplitudes (PAs) of the normalized power spectra of the EHG signals, along pregnancy (from early to later recorded records), derived from a variety of frequency bands, revealed that for the preterm group of records, in comparison to all other term delivery groups, the frequency spectrum of the frequency band B0L (0.08–0.3 Hz) shifts toward higher frequencies, and that the spectrum of the newly identified frequency band B0L’ (0.125–0.575 Hz), which approximately matches the Fast Wave Low band, becomes stronger. The most promising features to separate between the later preterm group and all other later term delivery groups appear to be MF (p=1.1⋅10−5) in the band B0L of the horizontal signal S3, and PA (p=2.4⋅10−8) in the band B0L’ (S3). Moreover, the PA in the band B0L’ (S3) showed the highest power to individually separate between the later preterm group and any other later term delivery group. Furthermore, the results suggest that in preterm pregnancies the resting maternal heart rate decreases between the 23rd and 31st week of gestation.

Optical properties of epitaxially grown GaN:Ge thin films

To better understand the scintillation properties of gallium nitride, a detailed characterization of Metal Organic Vapour Phase Epitaxy (MOVPE) grown GaN layers doped with Ge is conducted. This study includes the measurements of radioluminescence and photoluminescence spectra, thermally stimulated luminescence as well as scintillation decay kinetics. The representative radio- and photoluminescence spectra are composed of two bands. The narrow and fast band peaking at about 3.4 eV is ascribed to excitons and the broad and slow defect-related band with the maximum at about 2.2 eV is produced by carbon occupying nitrogen site. It has been found that the decay kinetics and amplitude of the exciton and defect emission are sensitive to the Ge doping level. In particular, the exciton- and the defect-related luminescence become faster with increasing Ge content. Charge trapping processes were also affected by the Ge doping.

EEG Activation Does Not Differ in Simple and Complex Episodes of Disorders of Arousal: A Spectral Analysis Study.

PurposeDisorders of arousal (DoA) are characterized by incomplete awakening from NREM sleep, with the admixture of both deep sleep and wake EEG activity. Previous observations suggested that changes in EEG activity could be detected in the seconds preceding DoA episodes. The aims of this work were to characterize the topography of EEG spectral changes prior to DoA episodes and to investigate whether or not behavioral complexity could be predicted by changes in EEG immediately preceding behavioral onsets.Patients and MethodsWe collected 103 consecutive video-polysomnographic recordings of 53 DoA adult patients and classified all episodes as simple, rising and complex arousal movements. For each episode, a 5-second window preceding its motor onset (“pre-event”) and a 60-second window from 2 to 3 minutes before the episodes (“baseline”) were compared. Subsequently, a between-group comparison was performed for the pre-event of simpler versus the more complex episodes.ResultsSpectral analysis over 325 DoA episodes showed an absolute significant increase prior to DoA episodes in all frequency bands excluding sigma, which displayed the opposite effect. In normalized maps, the increase was relatively higher over the central/anterior areas for both slow and fast frequency bands. No significant differences emerged from the comparison between simpler and more complex episodes.ConclusionTaken together, these results show that deep sleep and wake-like EEG rhythms coexist over overlapping areas before DoA episodes, suggesting an alteration of local sleep mechanisms. Episodes of different complexity are preceded by a similar EEG activation, implying that they possibly share a similar pathophysiology.

Analysis of tomographic models using resolution and uncertainties: a surface wave example from the Pacific

SUMMARY Since most tomographic problems deal with imperfect data coverage and noisy data, an estimate of the seismic velocity in the Earth can only be a local average of the ‘true’ velocity with some attached uncertainty. We use the SOLA (subtractive optimally localized averages) method, a Backus–Gilbert-type method based on the resolution–uncertainty trade-off, to build a range of models of Rayleigh-wave velocities in the Pacific upper mantle. We choose one solution and show how to analyse the model using its resolution and uncertainties. We exploit the model statistics to evaluate the significance of deviations from a theoretical prediction: a half-space cooling model of the Pacific lithosphere. We investigate a slow-velocity anomaly located northeast of Hawaii, at about 200 km depth, and a pattern of alternatively slow- and fast-velocity bands, aligned approximately northwest to southeast, between 200 and 300 km depth. According to our resolution and uncertainty analyses, both features seem to be resolved.

Fast band power measurement technique in Wi-Fi systems using spectrum analyzer

Fast band power measurement technique used for measuring band power in wireless fidelity (Wi-Fi) devices. The difficulty during testing the band power is burst mode transmission takes longer time to measure. A test jig or specialized tool required is costly, which allows continuous mode of transmission for power measurement. The delivered average power by a Wi-Fi device depends upon the data rate which can also vary during time. The procedure proposed hereby developed using a spectrum analyzer to measure band power in the burst mode, in order to ensure repeatable results and optimized to enable fast measurements at low cost, less time and above 90% accurate.

A CNN Framework With Slow-Fast Band Selection and Feature Fusion Grouping for Hyperspectral Image Change Detection

Change detection approaches can detect changed areas of the same scene at different times. Hyperspectral remote-sensing images contain large amounts of spectral information at high resolution. As hyperspectral datasets become abundant, more and more change detection technologies use hyperspectral images as raw data. Hyperspectral images suffer from band redundancy. There is an urgent need to improve the directionality of change of information features. To solve these problems, in this article, we propose a CNN framework involving slow-fast band selection (SFBS) and feature fusion grouping (SFBS-FFGNET) for hyperspectral image change detection. The main contributions of this article are as follows: 1) based on slow feature analysis (SFA), an SFBS method is proposed, which selects slow and fast feature bands to better extract changed and unchanged features, to more effectively separate changed and unchanged pixels; 2) we used a difference matrix to enrich the level of change information to provide more change characteristics for change detection; and 3) an FFG method was used to generate a more discriminative feature group, and the related loss function was designed. Experimental results on multiple real hyperspectral datasets showed that SFBS can reduce the operating load on the computer and improve the accuracy of change detection, and FFG can also improve the accuracy of change detection. In summary, SFBS-FFGNET is superior to most existing change detection methods.