No abstract

The interactions between different EEG frequency bands have been widely investigated in normal and pathologic brain activity. Phase-amplitude coupling (PAC) is one of the important forms of this interaction where the amplitude of higher frequency oscillations is modulated by the phase of lower frequency activity. Here, we studied the dynamic variations of PAC of high (gamma and ripple) and low (delta, theta, alpha, and beta) frequency bands in patients with focal epilepsy in different sleep stages during the interictal period, in an attempt to see if coupling is different in more or less epileptogenic regions. Sharp activities were excluded to avoid their effect on the PAC. The results revealed that the coupling intensity was generally the highest in stage N3 of sleep and the lowest in rapid eye movement sleep. We also compared the coupling strength in different regions [seizure onset zone (SOZ), exclusively irritative zone, and normal zone]. PAC between high and low frequency rhythms was found to be significantly stronger in the SOZ compared to normal regions. Also, the coupling was generally more elevated in spiking channels outside the SOZ than in normal regions. We also examined how the power in the delta band correlates to the PAC, and found a mild but statistically significant correlation between slower background activity in epileptic channels and the elevated coupling in these channels. The results suggest that an elevated PAC may reflect some fundamental abnormality, even after exclusion of sharp activities and even in the interictal period. PAC may therefore contribute to understanding the underlying dynamics of epileptogenic brain regions.

No abstract

No abstract

To evaluate the possibility of detecting fast ripples (FRs) on the surface EEG of patients with focal pharmacoresistant epilepsy, and to investigate the relationship between scalp FRs and localization of the seizure onset zone (SOZ). We included 10 patients undergoing combined surface-intracranial EEG with ≥10 spikes in the surface EEG during the first 30 consecutive minutes of N3 sleep. FRs (≥4 consecutive oscillations above 250 Hz with an amplitude clearly exceeding that of the background) on the surface EEG (F3-C3, C3-P3, Fz-Cz, Cz-Pz, F4-C4, C4-P4) were visually marked, and verified by two EEG experts. FRs were categorized as related to the SOZ, if localized in the brain lobe of the SOZ. Low-amplitude FRs with a rate of 0.09/min were found in 6/10 patients: two exhibited events related to the SOZ, three showed no relationship with the SOZ, and in one patient the SOZ was not identified. It may be possible to detect FRs with surface EEG using subdermal electrodes in patients with focal epilepsy. The relationship between surface FRs and the SOZ remains unclear. Future studies aiming at a higher spatial EEG coverage are needed to elucidate their significance.

Objective Low‐voltage fast activity (LVF) and low‐frequency high‐amplitude periodic spiking (PS) are the two most common seizure‐onset patterns in mesiotemporal lobe epilepsy, with different underlying mechanisms, pathology, and postsurgical outcome. The present work aims to investigate whether specific coupling patterns of high‐frequency oscillations (HFOs >80 Hz) and low‐frequency waves in the interictal period may distinguish these two patterns, and also seizure‐onset zone (SOZ) from non‐SOZ as a secondary aim. Methods We used intracranial electroencephalography (iEEG) data (during non–rapid eye movement [NREM] sleep) of 18 patients with either LVF or PS seizure‐onset patterns. We investigated the interaction between HFOs (ripples: 80‐250 Hz and fast ripples: >250 Hz) and slow oscillations (slow‐delta, delta, and theta waves). We compared classic features (amplitude, duration, frequency, and power) and phase of coupling between HFOs and slower oscillations inside and outside the SOZ. We then used these features to classify HFOs and subsequently patients into LVF and PS groups. Results Ripples in the LVF group had significantly longer duration, lower frequency, and higher amplitude than in the PS group. The phase of slow oscillations at which HFOs occur is different between the LVF and PS HFOs (LVF, mostly at the peak or the transition of peak to trough; PS, mostly during the transition of trough to peak). HFOs associated with theta waves best discriminate seizure‐onset patterns. The coupling phase improves the classification of HFOs and patients to either LVF or PS groups, and also the classification of HFOs in SOZ and non‐SOZ. Significance The phase of coupling of HFOs and low‐frequency waves may help to not only identify the SOZ, but also to classify patients with different types of seizure‐onset patterns. It likely reflects that different disease processes are involved in these patterns during the interictal period.

Fine-tuning a network which has been trained on a large dataset is an alternative to full training in order to overcome the problem of scarce and expensive data in medical applications. While the shallow layers of the network are usually kept unchanged, deeper layers are modified according to the new dataset. This approach may not work for ultrasound images due to their drastically different appearance. In this study, we investigated the effect of fine-tuning different layers of a U-Net which was trained on segmentation of natural images in breast ultrasound image segmentation. Tuning the contracting part and fixing the expanding part resulted in substantially better results compared to fixing the contracting part and tuning the expanding part. Furthermore, we showed that starting to finetune the U-Net from the shallow layers and gradually including more layers will lead to a better performance compared to fine-tuning the network from the deep layers moving back to shallow layers. We did not observe the same results on segmentation of X-ray images, which have different salient features compared to ultrasound, it may therefore be more appropriate to fine-tune the shallow layers rather than deep layers. Shallow layers learn lower level features (including speckle pattern, and probably the noise and artifact properties) which are critical in automatic segmentation in this modality.

No abstract