Browsing by Author "Taşdemir, Şakir"

Now showing 1 - 6 of 6

Designing a System That Records the Sleeping Position Data of Sleep Apnea Patients
(2022) Vatansev, Hülya; Vatansev, Hüsamettin; Gölcük, Adem; Taşdemir, Şakir; Balcı, Mehmet; Küçüktürk, Serkan
Sleeping positions have a significant impact on exposure of apnea patients to sleep apnea. In this study, the sleeping position of the patient was read with the STM microcontroller by using the body position sensor (SleepSense 1/8" Plug DC Body Position Sensor Kit) produced by the sleep sense company. This sensor produces results with analog signals between 0-2V. This analog signal was read using the ADC feature of the microcontroller. This signal read by the microcontroller was sent to the computer via the USB port. The C# software prepared on the computer reads the data from the microcontroller and saves this data and the arrival time of the data to the bit TXT file. The data in this TXT file is ready to be evaluated by signal processing methods. These data, together with other data obtained from the polysomnography device, can be used to learn the body position of the patient at the time of apnea.
Citation - WoS: 6
Citation - Scopus: 7
Improving Efficiency in Convolutional Neural Networks With 3d Image Filters
(Elsevier Sci Ltd, 2022) Uyar, Kübra; Taşdemir, Şakir; Ülker, Erkan; Ünlükal, Nejat; Solmaz, Merve
Background and objective: The effective performance of deep networks has provided the solution to various stateof-the-art problems. Convolutional Neural Network (CNN) is accepted as an accurate, effective, and reliable practice in image-based applications. However, there is a need to use pre-trained models in case of insufficient data in CNN. This study aims to present an alternative solution to this problem with the proposed 3D image based filter generation approach with simpler CNNs for the classification of small datasets. Methods: In this study, a novel 3D image filters-based CNN (Hist3DCNN) is proposed. The proposed filter generation approach is based on 3D object images taken from different perspectives. The efficiency of Hist3DCNN is shown on a novel histological dataset that contains blood, connective, epithelium, muscle, and nerve tissue images. Various case studies are carried out with generated filters assigned as the initial value to AlexNet and the designed Hist3DCNN model that is simpler than AlexNet. Results: Based on results, the classification accuracy of AlexNet with proposed filters used in convolution layers were 84.65% and 85.34%. The accuracy was increased to 85.47% by Hist3DCNN on the histological image classification. Moreover, four different benchmark datasets were tested to demonstrate the robustness of Hist3DCNN on various datasets. Conclusions: This study provides a new aspect to literature due to 3D image-based filter generation approach to initialize convolution filters. Experimental results validate that Hist3DCNN can be used as a filter value initialization method with simple CNN models that contain less learnable parameters for the classification task of small datasets.
Citation - WoS: 8
Citation - Scopus: 10
Machine Learning-Based Detection of Sleep-Disordered Breathing Type Using Time and Time-Frequency Features
(Elsevier Sci Ltd, 2022) Balcı, Mehmet; Taşdemir, Şakir; Özmen, Güzin; Gölcük, Adem
Sleep-disordered breathing is a disease that many people experience unconsciously and can have very serious consequences that can result in death. Therefore, it is extremely important to analyze the data obtained from the patient during sleep. It has become inevitable to use computer technologies in the diagnosis or treatment of many diseases in the medical field. Especially, advanced software using artificial intelligence methods in the diagnosis and decision-making processes of physicians is becoming increasingly widespread. In this study, we aimed to classify the sleep-disordered breathing type by using machine learning techniques utilizing time and time- fre-quency domain features. We used Pressure Flow, ECG, Pressure Snore, SpO2, Pulse and Thorax data from among the polysomnography records of 19 patients. We employed digital signal processing methods for six types of physiological data and obtained a total of 35 features using different feature extraction methods for five different classes (Normal, Hypopnea, Obstructive Apnea, Mixed Apnea, Central Apnea). Finally, we applied machine learning algorithms (Artificial Neural Network, Support Vector Machine, Random Forest, Naive Bayes, K Nearest Neighborhood, Decision Tree and Logistic Regression) on 5-class and 35-feature data sets. We used10 fold cross validation to verify the classification success. Our main contribution to the literature is that we developed a classification system to score all four different types of sleep-disordered breathing simultaneously by using 6 types of PSG data. As a five-class scoring result, the Random Forest (RF) algorithm showed the highest success with 76.3 % classification accuracy. When Hypopnea was excluded from the evaluation, classification accuracy increased to 86.6% for three Apnea-type disorders. Our proposed method provided 89.7% accuracy for the diagnosis of Obstructive Apnea by the RF classifier. The results show that time and time-frequency domain features are distinctive in Sleep-disordered breathing scoring, which is a very difficult process for physicians and a diagnostic support system can be design by evaluating many PSG data simultaneously.
Citation - WoS: 10
Citation - Scopus: 15
Multi-Class Brain Normality and Abnormality Diagnosis Using Modified Faster R-Cnn
(ELSEVIER IRELAND LTD, 2021) Uyar, Kübra; Taşdemir, Şakir; Ülker, Erkan; Öztürk, Mehmet; Kasap, Hüseyin
Background and Objective: The detection and analysis of brain disorders through medical imaging techniques are extremely important to get treatment on time and sustain a healthy lifestyle. Disorders cause permanent brain damage and alleviate the lifespan. Moreover, the classification of large volumes of medical image data manually by medicine experts is tiring, time-consuming, and prone to errors. This study aims to diagnose brain normality and abnormalities using a novel ResNet50 modified Faster Regions with Convolutional Neural Network(R-CNN) model. The classification task is performed into multiple classes which are hemorrhage, hydrocephalus, and normal. The proposed model both determines the borders of the normal/abnormal parts and classifies them with the highest accuracy. Methods: To provide a comprehensive performance analysis in the classification problem, Machine Learning(ML) and Deep Learning(DL) techniques were discussed. Artificial Neural Network(ANN), AdaBoost(AB), Decision Tree(DT), Logistic Regression(LR), Naive Bayes(NB), Random Forest(RF), and Support Vector Machine(SVM) were used as ML models. Besides, various Convolutional Neural Network(CNN) models and proposed ResNet50 modified Faster R-CNN model were used as DL models. Methods were validated using a novel brain dataset that contains both normal and abnormal images. Results: Based on results, LR obtained the highest result among ML methods and DenseNet201 obtained the highest results among CNN models with the accuracy of 84.80% and 85.68% for the classification task, respectively. Besides, the accuracy obtained by the proposed model is 99.75%. Conclusions: Experimental results demonstrate that the proposed model has yielded better performance for detection and classification tasks. This artificial intelligence(AI) framework can be utilized as a computer-aided medical decision support system for medical experts.
Citation - WoS: 3
Citation - Scopus: 4
Optimization of Electricity Generation Parameters With Microbial Fuel Cell Using the Response Surface Method
(Springer Heidelberg, 2022) Neşeli, Süleyman; Dinçer, Kevser; Taşdemir, Şakir; Hayder, Mustafa Akram
Due to an ever-increasing population and developments in technology, the demand for energy has been increasing daily. In order to meet this demand, renewable and alternative energy sources that do not harm the environment are often recommended. One of these alternative sources can be obtained from human wastewater and is known as the Microbial Fuel Cell (MFC). This study aimed to estimate the electricity production performance of MFC, using Nafion 117 membrane with 10 x 10 and 11 x 11 cm(2) area by both active and sediment sludges. The responses obtained from 16 physical experiments performed according to the L-16(2(4) x 2(2)) Taguchi orthogonal index were analyzed by the response surface method (RSM). According to the analysis, it was determined that the quadratic polynomial equations created for the estimation of the reaction values of the active and sediment sludge had validity over 90%. The ANOVA analysis applied to determine parameter ethicalities also showed that the most effective parameter on the reactions from the sludge types considered was the resistance at different rates. Thus, the results suggest that the orthogonal design-based RSM model is an effective tool that provides key information for optimizing energy harvesting from MFC technology and saves time concerning experimental work.
Wavelet Dalgacık Dönüşümü ile Tıkayıcı Uyku Apnesi Tahmini ve Epok Sürelerinin Etkisi
(2021) Balcı, Mehmet; Gölcük, Adem; Küççüktürk, Serkan; Taşdemir, Şakir; Vatansev, Hüsamettin; Vatansev, Hülya
Tıkayıcı uyku apnesi halk arasında uykuda nefes durması olarak da bilinen çok ciddi bir halk sağlığı sorunudur. Bu sağlık sorununun tespit edilmesi ciddi laboratuvar tetkikleri gerektirmektedir. Polisomnografi (PSG) olarak adlandırılan bu tetkik sisteminde hastadan gece boyunca birçok fizyolojik veri toplanarak kaydedilir. Daha sonra bu veriler incelenerek teshis için kullanılır. Bu çalışmada yaşları 34 ile 73 arasında ve vücut kitle endeksleri 24,6 ile 49,3 arasında değişen 24 hastadan elde edilen gerçek veriler kullanılmıştır. Bu hastaların 17’si ciddi, 6’sı orta, 1’i de hafif derecede uyku apnesi teşhisi koyulmuş bireylerdir. 24 hastanın hastanenin uyku servisinde uyuma ve veri toplama için geçirdiği süre ortalama 5 saat 8 dakika 3 saniyedir. Bu çalışmada PSG ile toplanan fizyolojik verilerden olan pressure flow, pressuse snore ve thorax sinyalleri kullanılmıştır. Bu sinyaller önce epoklara ayrılmış, daha sonra ön işlemlerden geçirilmiştir. Farklı epok sürelerinin kullanıldığı çalışmada, her sinyalden wavelet dalgacık dönüşümü yöntemi ile sinyal özellikleri çıkarılarak bir özellikler veri seti oluşturulmuştur. Oluşturulan bu veri seti kullanılarak hastanın uyku sırasında meydana gelecek apnelerin önceden tahmin edilmesi amacıyla bir sistem geliştirilmiştir. Farklı sınıfandırıcıların da kullanıldığı bu sistemde ham sinyallerin bölümlendirilmesinde kullanılan epok sürelerin tahmin başarısına etkisi araştırılmıştır. Epok süresi 30 saniye olarak belirlendiğinde %88 doğruluk oranı elde edilirken, epok süresi 15 saniye olarak belirlendiğinde tahmin doğruluğu %93,3 olarak hesaplanmıştır. Epok süresi 5 saniye olarak belirlendiğinde ise tahmin başarısı %97,2 olarak gerçekleşmiştir. Sonuçlar, epok sürelerinin kısaltılmasının tahmin başarısını artırdığını göstermektedir. Bunun nedeni olarak apne olayının meydana geldiği ana daha yakın bir zaman diliminde elde edilen fizyolojik verilerin, meydana gelecek apneyi daha iyi tanımlamasıdır.