Browsing by Author "Yilmaz, Bulent"

Now showing 1 - 20 of 44

Automated quantification of immunomagnetic beads and leukemia cells from optical microscope images
(ELSEVIER SCI LTD, THE BOULEVARD, LANGFORD LANE, KIDLINGTON, OXFORD OX5 1GB, OXON, ENGLAND, 2019) Uslu, Fatma; Icoz, Kutay); Tasdemir, Kasim; Yilmaz, Bulent; 0000-0003-4542-2728; 0000-0002-0947-6166; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
Quantification of tumor cells is crucial for early detection and monitoring the progress of cancer. Several methods have been developed for detecting tumor cells. However, automated quantification of cells in the presence of immunomagnetic beads has not been studied. In this study, we developed computer vision based algorithms to quantify the leukemia cells captured and separated by micron size immunomagnetic beads. Color, size based object identification and machine learning based methods were implemented to quantify targets in the images recorded by a bright field microscope. Images acquired by a 40x or a 20x objective were analyzed, the immunomagnetic beads were detected with an error rate of 0.0171 and 0.0384 respectively. Our results reveal that the proposed method attains 91.6% precision for the 40x objective and 79.7% for the 20x objective. This algorithm has the potential to be the signal readout mechanism of a biochip for cell detection. (C) 2019 Elsevier Ltd. All rights reserved.
Automatic body part and pose detection in medical infrared thermal images
(TAYLOR & FRANCIS LTD2-4 PARK SQUARE, MILTON PARK, ABINGDON OR14 4RN, OXON, ENGLAND, 2021) Ozdil, Ahmet; Yilmaz, Bulent; 0000-0002-6651-1968; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Ozdil, Ahmet; Yilmaz, Bulent
Automatisation and standardisation of the diagnosis process in medical infrared thermal imaging (MITI) is crucial because the number of medical experts in this area is highly limited.The current studies generally need manual intervention. One of the manual operations requires physician's determination of the body part and orientation. In this study automatic pose and body part detection on medical thermal images is investigated. The database (957 thermal images - 59 patients) was divided into four classes upper-lower body parts with back-front views. First, histogram equalization (HE) method was applied on the pixels only within the body determined using Otsu'sthresholding approach. Secondly, DarkNet-19 architecture was used for feature extraction, and principal component analysis (PCA) and t-distributed stochastic neighbour embedding (t-SNE) approaches for feature selection. Finally, the performances of various machine learning based classification methods were examined. Upper vs. lower body parts and back vs. front of upper body were classified with 100% accuracy, and back vs. front classification of lower body part success rate was 93.38%. This approach will improve the automatisation process of thermal images to group them for comparing one image with the others and to perform queries on the labeled images in a more user-friendly manner.
Comparison of deep learning and conventional machine learning methods for classification of colon polyp types
(SCIENDOBOGUMILA ZUGA 32A, WARSAW, MAZOVIA, POLAND, 2021) Dogan, Refika Sultan; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Dogan, Refika Sultan; Yilmaz, Bulent
Determination of polyp types requires tissue biopsy during colonoscopy and then histopathological examination of the microscopic images which tremendously time-consuming and costly. The first aim of this study was to design a computer-aided diagnosis system to classify polyp types using colonoscopy images (optical biopsy) without the need for tissue biopsy. For this purpose, two different approaches were designed based on conventional machine learning (ML) and deep learning. Firstly, classification was performed using random forest approach by means of the features obtained from the histogram of gradients descriptor. Secondly, simple convolutional neural networks (CNN) based architecture was built to train with the colonoscopy images containing colon polyps. The performances of these approaches on two (adenoma & serrated vs. hyperplastic) or three (adenoma vs. hyperplastic vs. serrated) category classifications were investigated. Furthermore, the effect of imaging modality on the classification was also examined using white-light and narrow band imaging systems. The performance of these approaches was compared with the results obtained by 3 novice and 4 expert doctors. Two-category classification results showed that conventional ML approach achieved significantly better than the simple CNN based approach did in both narrow band and white-light imaging modalities. The accuracy reached almost 95% for white-light imaging. This performance surpassed the correct classification rate of all 7 doctors. Additionally, the second task (three-category) results indicated that the simple CNN architecture outperformed both conventional ML based approaches and the doctors. This study shows the feasibility of using conventional machine learning or deep learning based approaches in automatic classification of colon types on colonoscopy images.
Comparison of Lung Tumor Segmentation Methods on PET Images
(IEEE, 2015) Eset, Kubra; Icer, Semra; Karacavus, Seyhan; Yilmaz, Bulent; Kayaalti, Omer; Ayyildiz, Oguzhan; Kaya, Eser; 0000-0002-8473-9720; 0000-0003-2954-1217; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yilmaz, Bulent; Ayyildiz, Oguzhan
Akciğer kanseri, tüm dünyada kansere bağlı gerçekleşen ölümlerin en sık nedenidir. Son zamanlarda, tümör içi 18Fflorodeoksiglukoz (FDG)’un tutulumunun düzgünlük, pürüzlülük ve düzenliliğini (yani tekstür özelliklerini) tanımlamak için PET görüntüleri üzerinde çeşitli görüntü işleme yaklaşımları kullanılmaktadır. Bunun ilk ve önemli aşaması tümörlü bölgenin diğer bölgelerden başarıyla ayrıştırılması, yani segmentasyonudur. Bu çalışmada, 36 hastadan alınan tek veya çok kesit görüntüler üzerinde kortalamalar, aktif kontur (yılan), Otsu eşikleme yaklaşımlarını kullanarak elde edilmiş alan ve hacimlerin ekibimizdeki nükleer tıp uzmanı tarafından değerlendirmesiyle karşılaştırması yapılmıştır. Sonuç olarak, Otsu eşikleme algoritmasının daha seçici davrandığı gözlenmiştir
Design and multichannel electromyography system-based neural network control of a low-cost myoelectric prosthesis hand
(Copernicus GmbH, 2021) Siddiq Ahmed, Saygin; Almusawi, Ahmed R. J.; Yilmaz, Bulent; Dogru, Nuran; 0000-0003-2954-1217; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yilmaz, Bulent
This study introduces a new control method for electromyography (EMG) in a prosthetic hand application with a practical design of the whole system. The hand is controlled by a motor (which regulates a significant part of the hand movement) and a microcontroller board, which is responsible for receiving and analyzing signals acquired by a Myoware muscle device. The Myoware device accepts muscle signals and sends them to the controller. The controller interprets the received signals based on the designed artificial neural network. In this design, the muscle signals are read and saved in a MATLAB system file. After neural network program processing by MATLAB, they are then applied online to the prosthetic hand. The obtained signal, i.e., electromyogram, is programmed to control the motion of the prosthetic hand with similar behavior to a real human hand. The designed system is tested on seven individuals at Gaziantep University. Due to the sufficient signal of the Mayo armband compared to Myoware sensors, Mayo armband muscle is applied in the proposed system. The discussed results have been shown to be satisfactory in the final proposed system. This system was a feasible, useful, and cost-effective solution for the handless or amputated individuals. They have used the system in their day-to-day activities that allowed them to move freely, easily, and comfortably.
Detection of Epileptic Seizures with Tangent Space Mapping Features of EEG Signals
(Institute of Electrical and Electronics Engineers Inc., 2021) Altındiş, Fatih; Yılmaz, Bülent; 0000-0003-2954-1217; 0000-0002-3891-935X; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Altindis, Fatih; Yilmaz, Bulent
Detection of epileptic seizures from EEG signals is well-studied topic for the last couple of decades. Lately, automated signal processing and machine learning methods were developed to detect epileptic seizures. However, most of the methods are tailored to subjects and require fine tuning of many parameters. In this study, we proposed to use Riemannian geometry-based signal processing method that already showed superior performance on brain-computer interface problems, to extract features. We showed that tangent space mapping features of EEG signals can be used to detect seizures with high accuracy and precision.
Detection of Ulcerative Colitis From Colonoscopy Images
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2018) Kacmaz, Rukiye Nur; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
Ulcerative colitis (UC) is a disease in which inner surface of colon is inflamed. Ulcers and open scars on the colon are observed. The complaint in the flare period is the frequent bloody diarrhea. Complaints of people with UC increase and decrease periodically. Colonoscopy is the most preferred approach for the visualization of the gastrointestinal tract for the diagnosis and follow-up of related diseases, and UC in particular. The lack of experience of the colonoscopist, complicated locality of the lesion, and the rush in the colonoscopy suite to complete the procedure as soon as possible may cause mistakes in visual analysis. In this study, 200 colonoscopy images (100 normal, 100 UC) were used. The statistical features such as gray level variance, gray level local variance, normalized variance, histogram range, and entropy were extracted from the images, and a normalized 200x5 feature matrix was formed. The normal images and images with UC were discriminated using support vector machines and k-nearest neighbors. It should be noted that the extraction of only 5 features from the colonoscopy images resulted in 95% accuracy. This study demonstrated the feasibility of the development of software tools for aiding the physicians in the diagnosis of colon diseases.
Detection of Variation Instances on Colonoscopy Videos using Structural Similarity Index
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2018) Kacmaz, Rukiye Nur; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
The aim of this study is to reduce the number of images extracted from the videos recorded by the specialists during the colonoscopy process for further examination, thereby enabling the specialist to deal with fewer images. Since the images obtained from the videos are very similar, the main assumption of this study is that the whole video can be represented by fewer images. The approach used in this study is the structural similarity index. Totally, images were obtained from 4 different videos coming from healthy, ulcerative colitis, Crohn's, and polyp patients. The noisy images in these videos were eliminated manually. When the structural similarity index between two consecutive clear images was less than 0.83, the second image was selected and shown to the specialist for his/her examination. By this way, the frames carrying significantly new information from the videos were defined as the variation instances. The tests on healthy or diseased colon videos showed that only 5-10% of the clear images provide significantly new information.
Discrimination of Rest, Motor Imagery and Movement for Brain-Computer Interface Applications
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2018) Ozturk, Nedime; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
Brain-computer interface (BCI) is a system that provides a means to control prosthesis, wheelchair, or similar devices using brain waves without direct motor nervous system involvement. For this purpose, brain waves obtained from multiple electrodes placed on the scalp (EEG, Electroencephalogram) are used. Emotiv Epoc used to obtain EEG signals is a low-cost device and has real-time applications.. The aim of this study is the detection of rest, imagination and real movement using EEG signals obtained by Emotiv Epoc headset. As a result, As a result, the data obtained from 39 trials from a female subject were classified resting, motion imagination and movement, according to 97.4% accuracy by using the statistical features of distortion, logarithm energy entropy, energy, Shannon entropy and kurtosis.In this study, it has been shown that this system can be remarkably successful for BCI applications.
Effect of Bilinear Interpolation on the Texture Analysis of Colonoscopy Images
(IEEE345 E 47TH ST, NEW YORK, NY 10017 USA, 2017) Kacmaz, Rukiye Nur; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
Interpolation is a method that is used to obtain unknown intensities with the help of known intensities on an image. This method is frequently used in the literature to eliminate light reflection on colonoscopy images. Texture features are the most important characteristics used to describe the region or objects of interest in the image. They are the measures of intensity variation of a surface that determine properties such as smoothness, roughness, and regularity. The aim of this study is to find out the how bilinear interpolation applied on colonoscopy images with reflection impact texture features obtained from the same images. A research carried out to make reasonable comparison between a texture feature from an image with no reflection and the same feature obtained from the same image with synthetically added reflections with various percentages. Using the approaches like gray level co-occurence matrix (GLCM), gray level run length matrix (GLRLM), neighborhood gray tone difference matrix (NGTDM) 126 features were extracted from each 32x32 sub-images coming from 610 colonoscopy images. Several of the features extracted from sub-images with no reflection and reflection were not statistically significantly different, while majority of them were affected from the reflections.
Effect of interpolation on specular reflections in texture-based automatic colonic polyp detection
(WILEY, 111 RIVER ST, HOBOKEN 07030-5774, NJ USA, 2020) Kacmaz, Rukiye Nur; Yilmaz, Bulent; Aydin, Zafer; 0000-0002-3237-9997; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
Reflections of LED light cause unwanted noise effects called specular reflection (SR) on colonoscopic images. The aim of this study was to seek answers to the following two questions. (a) How are the texture features used in automatic detection of polyps affected by the interpolation on specular reflections? (b) If they are affected does it really affect the classification performance? In order to answer these questions, we used 610 colonoscopy images, and divided each image into tiles whose sizes were 32-by-32 pixels. From these tiles, we selected the ones without any specular reflection. We added different shape and size specular reflections cropped from real images onto the reflection-free tiles. We then used the nearest neighbors, bilinear and bicubic interpolation techniques on the tiles on which SRs were added. On these tiles we extracted 116 texture features using 3 second-order approaches, and 4 first-order statistics. First, we used paired samplettest. Second, we performed automatic classification of polyps and background using random forest and k nearest neighbors (k-NN) approaches using the texture features for different combinations of specular reflections added on the tiles from the polyp or background. The results showed that depending on the size of specular reflection, interpolation can cause a significant difference between the texture features that were coming from reflection-free tiles and the same tiles on which interpolation was performed. In addition, we note that bicubic interpolation may be preferred to eliminate specular reflection when texture features are used for background and polyp discrimination.
An effective colorectal polyp classification for histopathological images based on supervised contrastive learning
(ELSEVIER, 2024) Yengec-Tasdemir, Sena Busra; Aydin, Zafer; Akay, Ebru; Dogan, Serkan; Yilmaz, Bulent; 0000-0001-7686-6298; 0000-0003-2954-1217; AGÜ, Mühendislik Fakültesi, Bilgisayar Mühendisliği Bölümü; Aydin, Zafer; Yilmaz, Bulent
Early detection of colon adenomatous polyps is pivotal in reducing colon cancer risk. In this context, accurately distinguishing between adenomatous polyp subtypes, especially tubular and tubulovillous, from hyperplastic variants is crucial. This study introduces a cutting-edge computer-aided diagnosis system optimized for this task. Our system employs advanced Supervised Contrastive learning to ensure precise classification of colon histopathology images. Significantly, we have integrated the Big Transfer model, which has gained prominence for its exemplary adaptability to visual tasks in medical imaging. Our novel approach discerns between in-class and out-of-class images, thereby elevating its discriminatory power for polyp subtypes. We validated our system using two datasets: a specially curated one and the publicly accessible UniToPatho dataset. The results reveal that our model markedly surpasses traditional deep convolutional neural networks, registering classification accuracies of 87.1% and 70.3% for the custom and UniToPatho datasets, respectively. Such results emphasize the transformative potential of our model in polyp classification endeavors.
An effective colorectal polyp classification for histopathological images based on supervised contrastive learning
(Elsevier, 2024) Yengec-Tasdemir,Sena Busra; Aydin,Zafer; Akay,Ebru; Doğan,Serkan; Yilmaz,Bulent; 0000-0001-7686-6298; AGÜ, Fen Bilimleri Enstitüsü, Elektrik ve Bilgisayar Mühendisliği Ana Bilim Dalı; Aydın, Zafer; Yilmaz, Bulent
Early detection of colon adenomatous polyps is pivotal in reducing colon cancer risk. In this context, accurately distinguishing between adenomatous polyp subtypes, especially tubular and tubulovillous, from hyperplastic variants is crucial. This study introduces a cutting-edge computer-aided diagnosis system optimized for this task. Our system employs advanced Supervised Contrastive learning to ensure precise classification of colon histopathology images. Significantly, we have integrated the Big Transfer model, which has gained prominence for its exemplary adaptability to visual tasks in medical imaging. Our novel approach discerns between in-class and out-of-class images, thereby elevating its discriminatory power for polyp subtypes. We validated our system using two datasets: a specially curated one and the publicly accessible UniToPatho dataset. The results reveal that our model markedly surpasses traditional deep convolutional neural networks, registering classification accuracies of 87.1% and 70.3% for the custom and UniToPatho datasets, respectively. Such results emphasize the transformative potential of our model in polyp classification endeavors
Emotion Detection Using Multivariate Synchrosqueezing Transform via 2D Circumplex Model
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2018) Ozel, Pinar; Akan, Aydin; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
Emotion detection by utilizing signal processing methods is a challenging area. An open issue in emotional modeling is to obtain an optimum feature set to use for the classification process. This study proposes an approach for emotional state classification by the investigation of EEG signals via multivariate synchrosqueezing transform (MSST). MSST is a post -processing technique to compose a localized time -frequency representation yielding multivariate syncyrosqueezing coefficients. After obtaining these coefficients from EEG signals for 18 subjects from DEAP dataset, coefficients and self assessment -mannequins (SAM) labels of those subjects are used for emotional state classification by using support vector machines (SVM) nearest neighbor, decision tree, and ensemble methods. The accuracy rate is 70.6% for high valence high arousal (HVHA), 75.4% for low valence high arousal (LVHA), 77.8% for high valence low arousal (HVLA), and 77.2% for low valence low arousal (LVLA) cases using SVM.
Emotion Elicitation Analysis in Multi-Channel EEG Signals Using Multivariate Empirical Mode Decomposition and Discrete Wavelet Transform
(IEEE345 E 47TH ST, NEW YORK, NY 10017 USA, 2017) Ozel, Pinar; Akan, Aydin; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yilmaz, Bulent
In recent years, wavelet-based, Fourier-based and Hilbert-based time-frequency methods attracted attention in emotion state classification studies in human machine interaction. In particular, the Hilbert-based Empirical Mode Decomposition and Wavelet-based Discrete Wavelet Transform have found applications in emotional state analysis. In this study, a model of emotional elicitation is proposed in which the classification is made by using the features of the wavelet coefficients obtained after applying the Discrete Wavelet Transform to IMFs achieved by using Multivariate Empirical Mode Decomposition. Accordingly, EEG data available in the DEAP database were classified as low / high for valence, activation, and dominance dimensions, and 4 different classifiers were used in the classification phase. The best ratios of valence, activation and dominance were obtained ideally 70.1%, 58.8%, 60.3% respectively.
Emotion Recognition Classification in EEG Signals Using Multivariate Synchrosqueezing Transform
(IEEE345 E 47TH ST, NEW YORK, NY 10017 USA, 2017) Ozel, Pinar; Akan, Aydin; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yilmaz, Bulent
Electrophysiological data processing can take place both in time and in frequency domains as well as in the joint time-frequency domain. Short Time Fourier Transform and Wavelet Transform are commonly used time-frequency analysis methods. The limitations of these methods initiated the use of methods such as synchrosqueezing and multivariate synchrosqueezing methods. In our proposed method 88.9%, 77.8%, 80.6% accuracy rates were obtained respectively for the valence, activation and dominance parameters using and multivariate synchrosqueezing methods and support vector machines(SVM) which yields better results than most of the other methods mentioned in the literature.
Emotional State Analysis from EEG signals via Noise-Assisted Multivariate Empirical Mode Decomposition Method
(IEEE345 E 47TH ST, NEW YORK, NY 10017 USA, 2017) Ozel, Pinar; Akan, Aydin; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yilmaz, Bulent
Emotional state analysis is an interdisciplinary arena because of the many parameters that encompass the complex neural structure and electrical signals of the brain and in terms of emotional state differences. In recent years, emotional state data have been examined by using data-driven methods such as Empirical Mode Decomposition as well as classical time-frequency methods. Although Empirical Mode Decomposition has many advantages, it has disadvantages such as being designed for univariate data, prone to mode mixing, and providing signal via a sufficient number of the local extrema. To overcome these disadvantages, in this study, the Noise-Assisted Multivariate Empirical Mode Decomposition has been shown to classify the emotional state using electroencephalographic signals.
Emotional State Sensing by Using Hybrid Multivariate Empirical Mode Decomposition and Synchrosqueezing Transform
(IEEE, 345 E 47TH ST, NEW YORK, NY 10017 USA, 2018) Ozel, Pinar; Akan, Aydin; Yilmaz, Bulent; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü
In recent years, utilizing Hilbert-based time frequency methods in emotional state sensing research attracted attention in the brain computer interfaces. Primarily, Hilbert Transform-based empirical mode decomposition (EMD) was found to be suitable for emotional state modeling studies. In more recent studies, models of emotional state recognition were proposed in which the classification was implemented by using the features obtained after applying the time, frequency, and time frequency domain methods to intrinsic mode functions achieved by operating EMD. In this study, an analysis of emotional state recognition is proposed by using the features of the synchrosqueezing coefficients obtained in the classification process after applying the Synchrosqueezing Transform to intrinsic mode functions achieved by using Multivariate EMD. As a result, EEG data available in the DEAP database were categorized as low and high for valence, activation, and dominance dimensions, and 4 different classifiers were utilized in the classification process. The most satisfying ratios of valence, activation and dominance were attained 76%, 68%, and 68% respectively.
An FDTD-based computer simulation platform for shock wave propagation in electrohydraulic lithotripsy
(ELSEVIER, 2013) Yilmaz, Bulent; Ciftci, Emre; 0000-0003-2954-1217; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Yilmaz, Bulent
Extracorporeal Shock Wave Lithotripsy (ESWL) is based on disintegration of the kidney stone by delivering high-energy shock waves that are created outside the body and transmitted through the skin and body tissues. Nowadays high-energy shock waves are also used in orthopedic operations and investigated to be used in the treatment of myocardial infarction and cancer. Because of these new application areas novel lithotriptor designs are needed for different kinds of treatment strategies. In this study our aim was to develop a versatile computer simulation environment which would give the device designers working on various medical applications that use shock wave principle a substantial amount of flexibility while testing the effects of new parameters such as reflector size, material properties of the medium, water temperature, and different clinical scenarios. For this purpose, we created a finite-difference time-domain (FDTD)-based computational model in which most of the physical system parameters were defined as an input and/or as a variable in the simulations. We constructed a realistic computational model of a commercial electrohydraulic lithotriptor and optimized our simulation program using the results that were obtained by the manufacturer in an experimental setup. We, then, compared the simulation results with the results from an experimental setup in which oxygen level in water was varied. Finally, we studied the effects of changing the input parameters like ellipsoid size and material,temperature change in the wave propagation media, and shock wave source point misalignment. The simulation results were consistent with the experimental results and expected effects of variation in physical parameters of the system. The results of this study encourage further investigation and provide adequate evidence that the numerical modeling of a shock wave therapy system is feasible and can provide a practical means to test novel ideas in new device design procedures.
Feature Extraction and Classification in A Two-State Brain-Computer Interface
(IEEE, 2015) Altindis, Fatih; Yilmaz, Bulent; 0000-0002-3891-935X; 0000-0003-2954-1217; AGÜ, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü; Altindis, Fatih; Yilmaz, Bulent
Brain Computer Interface (BCI) technology is used to help patients who do not have control over motor neurons such as ALS or paralyzed patients, to communicate with outer world. This work aims to classify motor imageries using real-time EEG dataset, which was published by Graz University, Austria. The dataset consists of two-channel EEG signals of right-hand movement imagery and left-hand movement imagery of 8 subjects. There are a total of 120 motor imagery trials (60 left and 60 right) EEG signals recorded from each subject. EEG signals are filtered and feature vectors were extracted that consist of 24, 32 and 40 relative band power values (RBPV). In this work, feature vectors classified by three different methods, linear discriminant analysis (LDA), K nearest neighbor (KNN) and support vector machines (SVM). Results show that best performance was achieved by 24 RBPV feature vector and LDA classification method.