Dilek Göksel Duru

Dr. Dilek Göksel Duru

orcid

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 April 2020 in Bilişim Teknolojileri Dergisi
Verified
Journal Article
Closed access
PDF file
SMSamet Mete
Samet Mete
OÇOğuz Çakir
Oğuz Çakir
OBOguz Bayat
Oguz Bayat
DDDilek Göksel Duru
Dilek Göksel Duru
ADAdil Deniz Duru
Adil Deniz Duru
Tr en İnsanlardaki duygu durumu otonom sinir sistemi tarafından kontrol edilmektedir. Bu sebeple, olumlu veya olumsuz bir uyaran ile karşılaştığında otonom sinir sistemi çok kısa bir süre içerisinde, uyaran çeşidinin bireyde tetiklediği duygu türüne göre çeşitli bedensel farklılıklara sebebiyet vermektedir. Bedensel bu farklılıklardan bir tanesi de kişilerin gözbebeğinin uyaran çeşitine göre gösterdiği fizyolojik farklılıklardır. Yapılan araştırmalar göz bebeği hareketlerinin ve boyutunun ölçülmesinin yararlı bir girdi sinyali olabileceğini göstermektedir. İnsanlar olumsuz bir uyaran gördüğünde gözdeki pupil boyutunda genişleme, olumlu bir uyaran gördüğünde ise pupil boyutunda daralma oluşmaktadır. Bu bilgiler ışığında çalışma kapsamında, erkek ve kadın katılımcılara uygulanan, türlü kategorilerden oluşan büyük bir fotoğraflar dizisi olan IAPS içerisinden, katılımcılarda fazlasıyla zıt duygulanımlar meydana getiren uyaran sınıfları değerlik puanlarına göre tercih edilmiş ve uyaranlar olumlu, olumsuz ve nötr olmak üzere üç sınıfa ayrılmıştır. Çalışma sırasında IAPS’ten seçilmiş olan toplamda 60 adet fotoğraf kullanılmış ve 13 adet katılımcıya sunulmuş ve göz takip cihazı kullanılarak katılımcıların göz verileri veri tabanına kaydedilmiştir. Sol ve sağ pupil büyüklükleri ve fiksasyon süresi sınıflama için girdi olarak kullanılmıştır. Üç sınıf kullanılararak, kNN, Naive Bayes, Destek Vektör Makinaları, Doğrusal Diskriminant analizi, karar ağacı ve lojistik regresyon teknikleri uygulanmıştır. Düşük sınıflandırma başarısından ötürü, işlem sadece pozitif ve negatif sınıflar için tekrar hesaplanmıştır. Bu iki emosyonel durum için %68’lik bir oran ile k-NN sınıflandırma yönteminde en yüksek sınıflandırma başarısına ulaşılmıştır. Naive Bayesçi Sınıflandırıcı ve DVM %55, LDA %50, karar ağaçları ve lojistik regresyon %48’lik başarıya ulaşmıştır. Kişilerin çeşitli uyaranlara verdiği emosyonel yanıtların göz hareketlerine yansıyabileceği ve göz hareketlerinden kişinin emosyonel uyarılma düzeyi hakkında fikir sahibi olunabileceği düşünülmektedir. Emotional state is controlled by the autonomous nervous system (ANS). Thus, in the presence of a positive or negative type of a stimulus, ANS responses occur in a short time which can be observed as various physical output, depending on the type of emotion triggered by the stimulus type in the individual. One of these physical differences related to the type of stimulus is the pupil size variation and can be named as a physiological change to examine one’s emotional state. According to previous studies, pupil size and eye movement measurements were shown to be a useful input signal. Relying on that, emotion recognition by extracting eye gaze pattern is aimed in the present study. When a negative type of a stimulus triggers a person, pupil size seems to dilate. On the other hand, in the presence of a positive type of stimulus, pupil size is tightened. Based on this information, in the concept of this study, stimuli are applied to male and female volunteers. Stimuli, a total of 60 pictures, are selected from the IAPS image database, where different emotional stimuli sets are chosen concerning valence scores to form positive, neutral and negative stimulus classes. Thirteen volunteers participated in the study to perform the test paradigm and to attend the eye tracker measurement. Left and right pupil size values and fixation time parameters are used for classification purposes. The input features are classified for three classes using kNN, Naive Bayes, Support Vector Machine, Linear Discriminant Analysis, decision tree and logistic regression techniques. Low classification accuracy yields us to apply classification based on positive and negative stimuli. Analysis results demonstrated the best success rate with 68% for kNN algorithm for classification within these two emotion groups, where the application of Naïve Bayes and SVM results in a success rate of 55%, Linear Discriminant Analysis 50%, decision tree and logistic regression 48%. To conclude, eye movements can reflect the emotional responses of the subject and also predictions of the arousal level of the subjects might be performed.

16 July 2021 in Physical and Engineering Sciences in Medicine
Verified
Journal Article
Closed access
PDF file
DDDilek Göksel Duru
Dilek Göksel Duru
MAMay Alobaidi
May Alobaidi

23 September 2021 in Applied Bionics and Biomechanics
Verified
Journal Article
Open access
PDF file
SGSelen Güney
Selen Güney
SEMA ARSLAN KABASAKALSEMA ARSLAN KABASAKAL
SEMA ARSLAN KABASAKAL
ADAdil Deniz Duru
Adil Deniz Duru
DDDilek Göksel Duru
Dilek Göksel Duru
Although food consumption is one of the most basic human behaviors, the factors underlying nutritional preferences are not yet clear. The use of classification algorithms can clarify the understanding of these factors. This study was aimed at measuring electrophysiological responses to food/nonfood stimuli and applying classification techniques to discriminate the responses using a single-sweep dataset. Twenty-one right-handed male athletes with body mass index (BMI) levels between 18.5% and 25% (mean age: ) participated in this study voluntarily. The participants were asked to focus on the food and nonfood images that were randomly presented on the monitor without performing any motor task, and EEG data have been collected using a 16-channel amplifier with a sampling rate of 1024 Hz. The SensoMotoric Instruments (SMI) iView XTM RED eye tracking technology was used simultaneously with the EEG to measure the participants’ attention to the presented stimuli. Three datasets were generated using the amplitude, time-frequency decomposition, and time-frequency connectivity metrics of P300 and LPP components to separate food and nonfood stimuli. We have implemented -nearest neighbor (kNN), support vector machine (SVM), Linear Discriminant Analysis (LDA), Logistic Regression (LR), Bayesian classifier, decision tree (DT), and Multilayer Perceptron (MLP) classifiers on these datasets. Finally, the response to food-related stimuli in the hunger state is discriminated from nonfood with an accuracy value close to 78% for each dataset. The results obtained in this study motivate us to employ classifier algorithms using the features obtained from single-trial measurements in amplitude and time-frequency space instead of applying more complex ones like connectivity metrics.

1 January 2025 in Research in Sports Science
Verified
Journal Article
Open access
RTRamazan Semih Teke
Ramazan Semih Teke
GTGamze Takmaz
Gamze Takmaz
RGRecep Göçmen
Recep Göçmen
DDDilek Göksel Duru
Dilek Göksel Duru
ESEmin Süel
Emin Süel
ADAdil Deniz Duru
Adil Deniz Duru
From a neuroscientific perspective, the basketball free throw task serves as an ideal model for understanding cogni tive processes such as decision-making, attention, and action preparation, all of which influence the success of a free throw. In this study, brain electrical activity was recorded from eight skilled basketball players using a 16-channel active electrode set with a sampling rate of 1000 Hz to investigate the neural patterns associated with free throw behavior. Simultaneously, movement data was captured using cameras operating at a 100 Hz sampling rate. After performing artifact rejection on the electroencephalogram, the data were segmented based on successful and un successful free throws. Subsequently, the power values of delta, theta, alpha, and beta frequency bands were com puted for each trial and averaged the data across channels to form feature sets representing frontal, fronto-central, central, and parieto-occipital regions. Paired sample t-tests and independent samples t-tests were then conducted to examine potential differences between the pre- and post-shooting periods for both successful and unsuccessful trials, as well as between the pre-shooting periods of successful and unsuccessful trials. Machine learning techniques, including 5-fold cross-validation, were used to classify the trials. The results from the statistical tests revealed band power changes consistent with motor preparation, notably mu rhythm desynchronization in the central regions. For single-trial electroencephalogram classification, motor movement was classified with an accuracy of approximately 75% using a support vector machine with a fine Gaussian kernel. However, classifying the success or failure of the throw using electroencephalogram band power achieved only 55% accuracy. Although the band power values showed promise for classifying movement epochs, further feature extraction techniques are needed to improve classification accuracy and gain a deeper understanding of the neural mecha nisms underlying free throw performance in basketball. Cite this article as: Teke, S. R., Takmaz, G., Göçmen, R., Göksel Duru, D., Süel, E., & Duru, A. D. (2025). Analysis of brain electrical signals during the execution of basketball free throws: A machine learning approach. Research in Sports Science, 15, 0031, doi: 10.5152/rss.2025.25031.

1 October 2015
Verified
Conference Paper
Closed access
EIElif Isikci
Elif Isikci
DDDilek Göksel Duru
Dilek Göksel Duru
In this study, a semi automatic segmentation of multiple sclerosis lesions in magnetic resonance (MR) images based on active contours is done. Based on the seed area selection by the user, the parametric active contours which rely on energy minimization, try to shrink and bend the snake resulting in boundary curve (capture region). The method is tested in MATLAB firstly on simulation data, and the precision of internal and external contour extraction is verified. Next, real brain MR images are investigated based on active contours (snakes) for segmentation of lesions. The method is discussed according the active contour implementation findings of the simulation and the real data.

6 February 2026 in Brain Sciences
Verified
Journal Article
Open access
PDF file
MÖMetin Kerem Öztürk
Metin Kerem Öztürk
DDDilek Göksel Duru
Dilek Göksel Duru
Objectives: Decoding neural patterns for RGB colors from electroencephalography (EEG) signals is an important step towards advancing the use of visual features as input for brain–computer interfaces (BCIs). This study aims to overcome challenges such as inter-subject variability and limited data availability by investigating whether transfer learning and signal augmentation can improve decoding performance. Methods: This research introduces an approach that combines transfer learning for cross-subject information transfer and data augmentation to increase representational diversity in order to improve RGB color classification from EEG data. Deep learning models, including CNN-based DeepConvNet (DCN) and Adaptive Temporal Convolutional Network (ATCNet) using the attention mechanism, were pre-trained on subjects with representative brain responses and fine-tuned on target subjects to parse individual differences. Signal augmentation techniques such as frequency slice recombination and Gaussian noise addition improved model generalization by enriching the training dataset. Results: The combined methodology yielded a classification accuracy of 83.5% for all subjects on the EEG dataset of 31 previously studied subjects. Conclusions: The improved accuracy and reduced variability underscore the effectiveness of transfer learning and signal augmentation in addressing data sparsity and variability, offering promising implications for EEG-based classification and BCI applications.