REVIEW | doi:10.20944/preprints202107.0028.v2
Subject: Mathematics & Computer Science, Algebra & Number Theory Keywords: Deep Learning; Depression; Electroencephalogram; EEG; Depressive Disorders; Systematic Literature Review
Online: 10 November 2021 (15:29:25 CET)
Depression is considered by WHO as the main contributor to global disability and it poses dangerous threats to approximately all aspects of human life, in particular public and private health. This mental disorder is usually characterized by considerable changes in feelings, routines, or thoughts. With respect to the fact that early diagnosis of this illness would be of critical importance ineffective treatment, some development has occurred in the purpose of depression detection. EEG signals reflect the working status of the human brain by which are considered the most proper tools for a depression diagnosis. Deep learning algorithms have the capacity of pattern discovery and extracting features from the raw data which is fed into them. Owing to this significant characteristic of deep learning, recently, these methods have intensely utilized in the diverse field of researches, specifically medicine and healthcare. Thereby, in this article, we aimed to review all papers concentrated on using deep learning to detect or predict depressive subjects with the help of EEG signals as input data. Regarding the adopted search method, we finally evaluated 22 articles between 2016 and 2021. This article which is organized according to the systematic literature review (SLR) method, provides complete summaries of all exploited studies and compares the noticeable aspects of them. Moreover, some statistical analysis performs to gain a depth perception of the general ideas of the latest researches in this area. A pattern of a five-step procedure was also established by which almost all reviewed articles fulfilled the goal of depression detection. Finally, open issues and challenges in this way of depression diagnosis or prediction and suggested works as the future directions discussed.
ARTICLE | doi:10.20944/preprints201810.0720.v1
Subject: Engineering, Electrical & Electronic Engineering Keywords: electroencephalogram; event-driven signal acquisition; activity selection; data compression; adaptive rate filtering
Online: 30 October 2018 (09:22:56 CET)
The segmentation and de-noising are basic operations, required in every signal processing and classification system. The classical segmentation and de-noising approaches are time-invariant. Consequently, it results in the post processing of an unnecessary information and causes an increase in the system processing activity and power consumption. In this context, an efficient event-driven segmentation and de-noising technique is proposed. It is founded on the principles of level crossing and activity selection. Therefore, it can adapt its sampling frequency, segmentation window length and position along with the filter order by analyzing the input signal local characteristics. As a result, the computational complexity and the power consumption of the proposed system is reduced compared to the counter ones. The suggested system performance is compared with the classical one. It is done for the case of a multi-channel Electroencephalogram (EEG) signals. Results show a noticeable compression gain with an effective adaptation of the de-noising filters order. It aptitudes a significant computational gain, transmission data rate reduction and power consumption reduction of the proposed technique, compared to the counter ones. It shows that the proposed solution is an attractive candidate to embed in the new generation EEG wearables.
ARTICLE | doi:10.20944/preprints202110.0375.v1
Subject: Engineering, Biomedical & Chemical Engineering Keywords: Brain-Computer Interface (BCI), Convolutional neural network (CNN), Electroencephalogram (EEG), Explainable artificial intelligence (XAI)
Online: 26 October 2021 (11:45:00 CEST)
Functional connectivity (FC) is a potential candidate that can increase the performance of brain-computer interfaces (BCIs) in the elderly because of its compensatory role in neural circuits. However, it is difficult to decode FC by current machine learning techniques because of a lack of its physiological understanding. To investigate the suitability of FC in BCI for the elderly, we propose the decoding of lower- and higher-order FCs using a convolutional neural network (CNN) in six cognitive-motor tasks. The layer-wise relevance propagation (LRP) method describes how age-related changes in FCs impact BCI applications for the elderly compared to younger adults. Seventeen younger (24.5±2.7 years) and twelve older (72.5±3.2 years) adults were recruited to perform tasks related to hand-force control with or without mental calculation. CNN yielded a six-class classification accuracy of 75.3% in the elderly, exceeding the 70.7% accuracy for the younger adults. In the elderly, the proposed method increases the classification accuracy by 88.3% compared to the filter-bank common spatial pattern (FBCSP). LRP results revealed that both lower- and higher-order FCs were dominantly overactivated in the prefrontal lobe depending on task type. These findings suggest a promising application of multi-order FC with deep learning on BCI systems for the elderly.
ARTICLE | doi:10.20944/preprints202201.0008.v1
Subject: Behavioral Sciences, Behavioral Neuroscience Keywords: functional near-infrared spectroscopy; electroencephalogram; cortico-cerebello-thalamo-cortical loop; transcranial electrical stimulation; transcranial magnetic stimulation
Online: 4 January 2022 (14:47:00 CET)
Background: Maladaptive neuroplasticity related learned response in substance use disorder (SUD) can be ameliorated using non-invasive brain stimulation (NIBS); however, inter-individual variability needs to be addressed for clinical translation. Objective: Our first objective was to develop a hypothesis for NIBS for learned response in SUD based on competing neurobehavioral decision systems model. Next objective was to conduct computational simulation of NIBS of cortico-cerebello-thalamo-cortical (CCTC) loop in cannabis use disorder (CUD) related dysfunctional “cue-reactivity” – a closely related construct of “craving” that is a core symptom. Our third objective was to test the feasibility of our neuroimaging guided rational NIBS approach in healthy humans. Methods: “Cue-reactivity” can be measured using behavioral paradigms and portable neuroimaging, including functional near-infrared spectroscopy (fNIRS) and electroencephalogram (EEG), metrics of sensorimotor gating. Therefore, we conducted computational simulation of NIBS, including transcranial direct current stimulation(tDCS) and transcranial alternating current stimulation(tACS) of the cerebellar cortex and deep cerebellar nuclei(DCN), of the CCTC loop for its postulated effects on fNIRS and EEG metrics. We also developed a rational neuroimaging guided NIBS approach for cerebellar lobule (VII) and prefrontal cortex based on healthy human study. Results: Simulation study of cerebellar tDCS induced gamma oscillations in the cerebral cortex while tTIS induced gamma-to-beta frequency shift. Experimental fNIRS study found that 2mA cerebellar tDCS evoked similar oxyhemoglobin(HbO) response in-the-range of 5x10-6M across cerebellum and PFC brain regions (=0.01); however, infra-slow (0.01–0.10 Hz) prefrontal cortex HbO driven(phase-amplitude-coupling, PAC) 4Hz, ±2mA (max.) cerebellar tACS evoked HbO in-the-range of 10-7M that was statistically different (=0.01) across those brain regions. Conclusion: Our healthy human study showed the feasibility of fNIRS of cerebellum and PFC as well as fNIRS-driven ctACS at 4Hz that may facilitate cerebellar cognitive function via the frontoparietal network. Future work needs to combine fNIRS with EEG for multi-modal imaging.