KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 6 of 6

Some of the metrics are blocked by your
consent settings
Decoding undergraduates’ English reading proficiency: EEG evidence and GPA–SES effects in a computerised English assessment
(2026-01-01)
Khemanuwong, Thapanee
;
Angsuwatanakul, Thanate
;
Phairot, Ekkapon
;
Mohamed Ismail, Shaik Abdul Malik
;
Bunyarit, Nagorn
This study investigates undergraduate students’ English reading proficiency through a sequential mixed-methods design that integrates neurocognitive evidence, behavioural performance, learner perceptions, and background characteristics within a computerised assessment context. In Phase 1, paper-based testing (PBT) and computer-based testing (CBT) were compared using electroencephalography (EEG) and qualitative inquiry. Five Thai EFL undergraduates completed parallel Thai Reading Evaluation and Decoding System (Thai-READS) tasks in both formats while EEG data were recorded using a single-channel MindWave Mobile 2 device. Wilcoxon signed-rank tests indicated no significant differences in Thai-READS scores between PBT and CBT (Z = −1.633, p = 0.102). EEG analyses revealed largely comparable cognitive processing across test modes; however, a significant difference in alpha frequency emerged for difficult items answered incorrectly (Z = −2.023, p = 0.043), with higher alpha frequency observed in the CBT condition, suggesting differential neural engagement under increased task difficulty. Semi-structured interviews further indicated that students generally perceived CBT as convenient, organised, and supportive of focus, while also expressing concerns related to technical reliability, digital navigation, and limited opportunities for annotation. Phase 2 analysed Thai-READS performance data from 3,274 undergraduates using a two-way analysis of variance to examine the effects of academic achievement (GPA) and socioeconomic status (SES). Results revealed significant main effects of both GPA and SES on reading proficiency (p < 0.05), with GPA demonstrating a larger effect size (η<sup>2</sup> = 0.06). The findings support the feasibility of CBT while highlighting the importance of considering cognitive processing and learner background to ensure valid and equitable assessment practices in Thai higher education.
Some of the metrics are blocked by your
consent settings
Memory Training in Dementia: A VR-BCI Prototype with N-Back and Brainwave Analysis
(2025-01-01)
Angthong, Sorawich
;
Angsuwatanakul, Thanate
;
Phairot, Ekkapon
;
Bunyasuwan, Meena
;
Khieukhajee, Jedsada
Dementia, a neurodegenerative condition leading to significant memory impairment in the elderly, currently lacks a curative treatment. Management strategies focus on mitigating symptom progression. This study presents the development and preliminary evaluation of a novel Virtual Reality (VR) game prototype integrated with Brain Computer Interface (BCI) technology designed to enhance memory, a key deficit in dementia, and potentially delay its progression. The VR game employed an adaptive N-back task for memory training, while a 4-channel OpenBCI system non-invasively monitored electroencephalography (EEG) signals from five healthy young adult volunteers (aged 18-22). Analysis of EEG power ratios revealed a trend towards increased alpha wave activity during memory encoding phases of the game. Specifically, during the Alpha EEG test, theta wave amplitudes were consistently higher across all channels (F3, F4, O1, O2) compared to alpha and beta waves. Furthermore, a descriptive trend indicated that theta wave amplitudes tended to increase with increasing difficulty in the Alpha EEG test. During the N-back tasks, theta wave amplitudes were observed to be descriptively higher during the memory encoding period compared to the recall period in frontal channels (F3, F4). The initial observations, while not statistically significant due to the limited sample size, support the feasibility of using the VR-BCI system to engage memory-related brain activity. To thoroughly evaluate the potential of this approach for cognitive rehabilitation and dementia mitigation, further comprehensive research involving larger and older cohorts utilizing longitudinal designs is warranted.
Some of the metrics are blocked by your
consent settings
Alleviating Chronic Stress in University Students: An EEG Analysis of Binaural Beat Therapy
(2025-01-01)
Aunchit, Sasithon
;
Angsuwatanakul, Thanate
;
Phairot, Ekkapon
;
Jinawan, Chulikon
;
Sirirattanapan, Boonrut
The COVID-19 pandemic has significantly elevated chronic stress levels, especially among university students. This study examined the effectiveness of binaural beat auditory stimulation in reducing chronic stress, using both the DASS-21 questionnaire and electroencephalography (EEG) as measurement tools. Six students identified with high stress were randomly assigned to one of three four-week interventions: alpha binaural beat music, beta binaural beat music, or non-binaural beat control music. EEG analysis revealed a general increase in power spectrum across all groups, indicating changes in brainwave activity. Notably, a significant increase in sample entropy—a measure of neural complexity—was observed exclusively in the alpha binaural beat group by week four. Correspondingly, this group showed the greatest reduction in self-reported DASS-21 stress scores. These findings suggest that alpha binaural beat stimulation promotes effective neural entrainment and is more successful than beta or non-binaural beat stimuli in alleviating chronic stress among university students. This study highlights the potential of binaural beat auditory stimulation as a non-invasive tool for improving mental health.
Some of the metrics are blocked by your
consent settings
Evaluating synthetic neuroimaging data augmentation for automatic brain tumour segmentation with a deep fully-convolutional network
(2024-06-01)
Asadi, Fawad
;
Angsuwatanakul, Thanate
;
O'Reilly, Jamie A.
Gliomas observed in medical images require expert neuro-radiologist evaluation for treatment planning and monitoring, motivating development of intelligent systems capable of automating aspects of tumour evaluation. Deep learning models for automatic image segmentation rely on the amount and quality of training data. In this study we developed a neuroimaging synthesis technique to augment data for training fully-convolutional networks (U-nets) to perform automatic glioma segmentation. We used StyleGAN2-ada to simultaneously generate fluid-attenuated inversion recovery (FLAIR) magnetic resonance images and corresponding glioma segmentation masks. Synthetic data were successively added to real training data (n = 2751) in fourteen rounds of 1000 and used to train U-nets that were evaluated on held-out validation (n = 590) and test sets (n = 588). U-nets were trained with and without geometric augmentation (translation, zoom and shear), and Dice coefficients were computed to evaluate segmentation performance. We also monitored the number of training iterations before stopping, total training time, and time per iteration to evaluate computational costs associated with training each U-net. Synthetic data augmentation yielded marginal improvements in Dice coefficients (validation set +0.0409, test set +0.0355), whereas geometric augmentation improved generalization (standard deviation between training, validation and test set performances of 0.01 with, and 0.04 without geometric augmentation). Based on the modest performance gains for automatic glioma segmentation we find it hard to justify the computational expense of developing a synthetic image generation pipeline. Future work may seek to optimize the efficiency of synthetic data generation for augmentation of neuroimaging data.
Some of the metrics are blocked by your
consent settings
Brain Function Analysis Using EEG Evidence: New Insights into English Paper-Based Versus Computer-Based Tests
(2024-01-01)
Khemanuwong, Thapanee
;
Angsuwatanakul, Thanate
;
Phairot, Ekkapon
;
Iramina, Keiji
;
Puttasakul, Tasawan
In today's disruptive era, where digital systems and the internet are central, assessment methods are transitioning from paper-based to computer-based tests. As digital technology becomes more accessible, there is growing interest in determining whether these new formats provide more effective means of evaluating learning outcomes. To investigate this, we examined performance and objective measures of brain function during both computer-based and paper-based reading comprehension tests. This study focuses on brain function analysis of computer-based and paper-based tests measured by EEG signals. Five healthy students at the Bl CEFR level voluntarily participated in two experimental conditions: paper-based testing (PBT) and computer-based testing (CBT). During the tests, EEG signals were recorded and analyzed using MATLAB to identify various features of brain activity. The results indicate that participants who performed better on paper-based tests showed greater familiarity with the test format. The power spectral density of EEG recordings, along with the average frequency of alpha and beta waves, were positively correlated with test familiarity. Specifically, the correlation coefficients were as follows: CBT-difficult (r=0.82, p < 0.05), CBT -easy (r=0.82,p < 0.05), PBT -difficult (r=0.82, p < 0.05), and PBT -easy (r=0.65,p < 0.05).
Some of the metrics are blocked by your
consent settings
Mind to Motion: EEG-Based Classification of Motor Imagery and Actual Hand Movements Using LSTM Models
(2023-01-01)
Chakkamallisery, Apoorva Sunil
;
Pelmo, Sonam Tenzin
;
Angsuwatanakul, Thanate
;
Pititheeraphab, Yutthana
;
Puttasakul, Tasawan
This study presents an Electroencephalogram (EEG) based classification model tailored to discern between motor imagery and real motor actions. Additionally, the study investigates the efficacy of employing Long Short-Term Memory (LSTM) deep learning models for EEG signal analysis. The EEG analysis comprised two distinct phases, aimed at validating the hypothesis of distinguishing motor action from motor imagery. The proposed LSTM-based classification model exhibited a notable accuracy of 62.5% in discriminating motor action from motor imagery, and a promising 72.5% accuracy in distinguishing between resting state and motor action. These findings highlight the potential of EEG-based approaches in motor-related applications, thus providing auspicious avenues for the future development of brain-computer interfaces (BCIs) and motor rehabilitation technologies.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results