Search

The most basic definition of the sleep is related with organism’s response to environmental stimuli. Sleep can be defined as a reversible, lessthan-awake response to the environment, which recurs on a daily cycle. All problems that affect sleep quality, duration and subject’s daily life are named as sleep disorder. Polysomnography is the gold standard for the evaluation of sleep signals. A polysomnography device records electroencephalography, electrooculography, and electromyography activities at the same time. A polysomnographic measurement can be extended with recording additiona lphysiological signals, like electrocardiography and body position, oxygen level in the blood, and snoring. These additional signals are important for the diagnosis of sleep disorders. The diagnosis of sleep disorder is done with analyzing laborious overnight polysomnography recording. Nowadays, to reduce duration and increase accuracy of diagnosis, artificial intelligence applications are used. These applications developed by using feature extraction-based machine learning or deep learning algorithms that generally apply to one or two polysomnography signals. By using artificial intelligence in sleep studies, duration of diagnosis reduces from hours to minutes and increases accuracy of diagnosis to over 90%. This study gives examples about artificial intelligence applications used in sleep studies.

Functional near-infrared spectroscopy (fNIRS) and its interaction with machine learning (ML) is a popular research topic for the diagnostic classification of clinical disorders due to the lack of robust and objective biomarkers. This review provides an overview of research on psychiatric diseases by using fNIRS and ML. Article search was carried out and 45 studies were evaluated by considering their sample sizes, used features, ML methodology, and reported accuracy. To our best knowledge, this is the first review that reports diagnostic ML applications using fNIRS. We found that there has been an increasing trend to perform ML applications on fNIRS-based biomarker research since 2010. The most studied populations are schizophrenia (n = 12), attention deficit and hyperactivity disorder (n = 7), and autism spectrum disorder (n = 6) are the most studied populations. There is a significant negative correlation between sample size (>21) and accuracy values. Support vector machine (SVM) and deep learning (DL) approaches were the most popular classifier approaches (SVM = 20) (DL = 10). Eight of these studies recruited a number of participants more than 100 for classification. Concentration changes in oxy-hemoglobin (Delta HbO) based features were used more than concentration changes in deoxy-hemoglobin (Delta b) based ones and the most popular Delta HbO-based features were mean Delta HbO (n = 11) and Delta HbO-based functional connections (n = 11). Using ML on fNIRS data might be a promising approach to reveal specific biomarkers for diagnostic classification., Institute of Photonic Sciences, We would like to thank Prof. Dr. Turgut Durduran from the Institute of Photonic Sciences (ICFO, Barcelona, Spain) for his valuable and constructive suggestions during the planning and development of this review.

Functional near-infrared spectroscopy (fNIRS) and its interaction with machine learning (ML) is a popular research topic for the diagnostic classification of clinical disorders due to the lack of robust and objective biomarkers. This review provides an overview of research on psychiatric diseases by using fNIRS and ML. Article search was carried out and 45 studies were evaluated by considering their sample sizes, used features, ML methodology, and reported accuracy. To our best knowledge, this is the first review that reports diagnostic ML applications using fNIRS. We found that there has been an increasing trend to perform ML applications on fNIRS-based biomarker research since 2010. The most studied populations are schizophrenia (n = 12), attention deficit and hyperactivity disorder (n = 7), and autism spectrum disorder (n = 6) are the most studied populations. There is a significant negative correlation between sample size (>21) and accuracy values. Support vector machine (SVM) and deep learning (DL) approaches were the most popular classifier approaches (SVM = 20) (DL = 10). Eight of these studies recruited a number of participants more than 100 for classification. Concentration changes in oxy-hemoglobin (Delta HbO) based features were used more than concentration changes in deoxy-hemoglobin (Delta b) based ones and the most popular Delta HbO-based features were mean Delta HbO (n = 11) and Delta HbO-based functional connections (n = 11). Using ML on fNIRS data might be a promising approach to reveal specific biomarkers for diagnostic classification., Institute of Photonic Sciences, We would like to thank Prof. Dr. Turgut Durduran from the Institute of Photonic Sciences (ICFO, Barcelona, Spain) for his valuable and constructive suggestions during the planning and development of this review.

In this study, it is aimed to investigate the analysis radiomics-machine learning on diagnostic performance in differential malign and benign breast lesions using mammography images. In this retrospective study included 101 patients (40 benign and 61 malign). 195 of region of interests (ROIs) were drawn manually by two expert radiologists. Then, using gray level thresholding and morphological operations techniques, each of ROI were segmented on “MATLAB 2020a” program. 126 radiomic features were extracted for each ROI. For eliminating redundant radiomics features, Kruskal Wallis and Relief feature selection methods were used respectively. A total 44 radiomics features were selected after feature selection process. Logistic regression, naive bayes, support vector machine and k-nearest neighbors machine learning algorithms (ML) were used to as classifiers. 10-fold cross validation was applied to measure and evaluate machine learning models. Accuracy, sensitivity and specificity were used as the primary measures of performance of radiomics-machine learning model. Among the machine learning algorithms, support vector machine had the best performance (93.3%, 95.6%, 91.1%). In addition, we found that the feature selection method improved the performance for all ML models. By building the radiomics-ML based analysis with the optimal feature subset, the performance of discrimination of benign and malign lesions showed excellent results which we believe would be useful for clinical practice.

Obstructive sleep apnea (OSA) is a common sleeping breathing disorder characterized by interruptions in breathing or obstructions in the airway. An early prediction of OSA may help in avoiding the disorder’s symptoms by identifying events before they happen. In this regard, we proposed a methodology for OSA prediction using both convolutional neural networks and traditional machine learning approaches. For this purpose, 30-second pre-apnea and non-apnea segments of electrocardiogram (ECG) recordings were extracted for various leading times and 2D scalogram images representing the time-frequency characteristics were generated from each segment. Deep features extracted from scalogram images using a modified residual network were fed separately into a support vector machine and two ensemble classifiers, namely random subspace k-nearest neighbors (kNN) and random subspace discriminant classifiers. The subspace kNN classifier outperformed other classifiers and achieved performance results up to 86,97% accuracy, 88,19% sensitivity, 84,26% specificity, and 85,70% positive predictive value. These results suggest that using machine learning approaches to classify deep features of single lead ECG scalogram images may improve prediction performance. Ultimately, the proposed method can be used as an useful approximation to identify impending OSA events

Currently, breast cancer affects many women worldwide. In recent years, many Computer-aided diagnosis (CAD) model have been developed for early diagnosis of breast cancer. An efficient CAD model is suggested to identify mammogram images as benign versus malignant in this study. The suggested CAD model constitutes four stages which are image acgusition, segmentation, feature extraction, feature selection and classification process. Gray level run matrix (GLRM) approach is used for feature extraction, while monarch butterfly optimization (MBO) for feature selection process. Support vector machine (SVM) algorithm is preferred for classification process. The suggested model has been tested on a private mammographic dataset. The suggested model (GLRM+MBO+SVM) shows an 0.944 of accuracy for breast lesion classification. Compared with similar studies, our proposed model showed good classification results for the breast lesion classification process.

Eklemeli/Katmanlı Üretim (Additive Manufacturing) olarak kabul edilen 3-B yazıcı teknolojisi, bilgisayar ortamında tasarlanan 3 boyutlu objeleri somut hale dönüştürebilen hızlı prototipleme araçlarıdır. Günümüzde 3-B yazıcı teknolojileri birçok alanda hayatı kolaylaştırdığı gibi, medikal alandaki uygulamaları da insan yaşamının kalitesini artırmış, en önemlisi yaşamlarını devam ettirmede büyük rol oynamıştır. 3-B yazıcı teknolojileri tıbbi implantların hızlı bir şekilde üretilmesini sağlamak ve doktorların ve cerrahların prosedürleri planlama yöntemlerini geliştirmek için büyük bir fırsat yaratmaktadır. Günümüzde hastalık, savaş, kaza ve yaralanma gibi sebeplerle eklem kayıpları çok görülmektedir. Bu eklem kayıplarını tedavi etmede kişinin hasarlı bölgesindeki parçanın çıkartılarak yerine geçebilecek kişinin dokusu ile uyumlu malzemeden üretilmiş medikal implantlar kullanılmaktadır. 3-B üretim teknolojisi, eklem kaybının yerine geçebilecek, kişiye özel ve kişinin dokusu ile uyumlu malzemeden üretilmiş implantların üretilmesine olanak vermektedir. Pazardaki standart implantların aksine kişinin durumuna özel implant üretilmesi ameliyatların risklerini azaltmanın yanı sıra cerrahların işini de kolaylaştırmaktadır. Titanyum alaşım ailesinden Ti-6Al-4V, dayanıklı ve hafif oluşları nedeniyle sternokostal implantlarda tercih edilmektedir. Bu çalışmada 3-B üretim teknolojilerinden toz esaslı eklemeli imalat yöntemi olan SLS (Seçimli Lazer Sinterleme) ile kişiye özel sternokostal eklem implantının üretim ve tasarım aşamaları incelenmiş, yeni bir implant tasarımı önerilmiştir. METÜM (Medikal Tasarım ve Üretim Merkezi, Sağlık Bilimleri Üniversitesi Gülhane Yerleşkesi, Ankara) tarafından oluşturulan ve hastaya implantasyonu gerçekleştirilen Ti-6Al-4V malzemeden üretilmiş bir sternokostal implantın analiz programında sonlu elemanlar analizi yapılmış, implanta binen yükler ve stres değerleri ölçülmüştür. Dünyada ve ülkemizde günümüze kadar tasarlanmış olan bazı sternokostal impla, 3-D printer technology, which is considered as Additive Manufacturing, is rapid prototyping tools that can transform 3D objects designed in computer environment into concrete. Nowadays, 3D printer technologies have made life easier in many areas, as well as applications in the medical field have increased the quality of human life, and most importantly, have played a major role in maintaining their lives. 3-D printer technologies create a great opportunity to enable the rapid production of medical implants and to improve the way doctors and surgeons plan procedures. Today, joint losses are common due to illness, war, accident, and injury. To treat these joint losses, medical implants made of materials that are compatible with the tissue of the person that can replace the part in the damaged area of the person are used. 3-D manufacturing technology allows the production of implants made of personal and compatible materials that can replace joint loss. Unlike the standard implants in the market, the production of implants specific to the person's condition not only reduces the risks of the surgeries but also facilitates the job of the surgeons. Ti-6Al-4V, a member of the titanium alloy family, is preferred in sternocostal implants due to its durability and lightness. In this study, the production and design stages of a custom sternocostal joint implant with SLS (Selective Laser Sintering), which is a powder-based additive manufacturing method from 3-D manufacturing technologies, was examined and a new implant design was proposed. In the analysis program of a sternocostal implant made of Ti-6Al4V material created by MDMC (Medical Design and Manufacturing Center, University of Health Sciences, Gülhane, Ankara) and implanted to the patient, finite element analysis was performed and the loads and stress values on the implant were measured. Some sternocostal implants designed to date in the world and in our country have been examined within the scope of the study, design d

Biyomedikal mühendisliği disiplinler arası bir çalışma sahasına sahiptir. Farklı alanlardan bir araya gelen araştırmacıların tek hedefi canlı hayatını kolaylaştırmak ve sağlıklı bir biçimde sürdürmesini sağlamaktır. Fizyolojik sinyallerin ve tıbbi görüntülerin işlenmesi biyomedikal mühendisliğinin popüler alanlarından biridir. Askeri alanlarda biyomedikal araştırmalar; askerlerin eğitimlerinin geliştirilmesi için kullanılan biyoelektronik ve yapay zeka çalışmalarını, muharebe sahasında askerlerin fizyolojik sinyallerinin izlenmesini, giyilebilir robotik sistemlerin geliştirilmesini, tramvaya bağlı hasarların tespit edilmesini ve yaralanmalarda insan-makine etkileşimini varıncaya kadar geniş bir alanı kapsamaktadır. Bu çalışmada ülkemizde ve dünyada askeri alanda yürütülen biyomedikal araştırmalar ile ilgili kısa bilgiler verilmiştir., Biomedical engineering has an interdisciplinary field of study. The only goal of researchers coming together from different fields is to make life easier and to maintain it in a healthy way. Processing physiological signals and medical images are o popular areas of biomedical engineering. biomedical research in military fields; It covers a wide area from bioelectronic, and artificial intelligence studies used for the development of soldiers' training, monitoring of soldiers' physiological signals on the battlefield, development of wearable robotic systems, detection of tram-related damages and human-machine interaction in injuries. This study gives brief information about biomedical research carried out in the military field in our country and the world.

Kalp hastalığı, yaygınlığı ve yüksek ölüm oranları nedeniyle insan sağlığını tehdit etmektedir. Kalp hastalığını tahmin etmek, geleneksel yöntemler kullanarak karmaşık bir iştir. Son yıllarda, kalp hastalıklarını tahmin etmek için makine öğrenimi teknikleri kullanılmaktadır. Bu çalışma kapsamında StatLog Kalp Hastalığı veri seti üzerinde Karar Ağacı, Naive Bayes, Lojistik Regresyon, Destek Vektör Makineleri ve K-En Yakın Komşu makine öğrenme teknikleri kullanılarak kalp hastalıklarının tespitine ilişkin ilişkin karşılaştırmalı bir analiz sunulmaktadır. Veri setindeki etkin öznitelikleri seçmek için Mann Whitney U testi kullanılmıştır. Makine öğrenimi algoritmalarının sınıflandırma performansı, doğruluk, kesinlik, duyarlılık ve F1-skoru açısından değerlendirilmiştir. Destek Vektör Makineleri 96.3% doğruluk, 95.83% kesinlik, 95.83% duyarlılık ve 95.83% F1skoru ile çalışmanın en iyi tahmin oranına sahip algoritması olmuştur. Bu çalışmanın klinisyenlere kalp hastalığını erken evrede tespit etmede yardımcı olacağına inanmaktayız., Heart disease threatens human health due to its prevalence and high mortality rates. Predicting heart disease is complicated task using traditional methods. In recent years, machine learning techniques have been utilized to predict heart diseases. In this study, a comparative analysis of heart disease detection is presented using Decision Tree, Naive Bayes, Logistic Regression, Support Vector Machines and K-Nearest Neighbor machine learning techniques on the StatLog Heart Disease dataset. Mann Whitney U test is utilized to select the influential features. The classification performance of machine learning algorithms was evaluated in terms of accuracy, precision, recall and F1 score. The Support Vector Machines was the algorithm with the best prediction rate of the study, with 96.3% accuracy, 95.83% precision, 95.83% sensitivity and 95.83% F1-score. We believe that this study will help clinicians detect heart disease at an early stage.

Temperature measurement is a vital part of daily newborn care. In temperature measurement, accurate measurements must be taken to detect deviations from expected values for the incubator and radiant heater function. Monitoring the temperature is keeping the baby in a thermoneutral environmental zone. The study tested body surface temperature under various clinical conditions using an affordable infrared thermography imaging technique. Temperature distributions are displayed as real-time videos and analyzed to evaluate average skin temperatures. This study demonstrates the design and implementation of a virtual temperature sensing application that can help neonatologists provide additional security to a newborn's skin temperature. The influence of different environmental conditions inside the radiant heater with respect to the surface temperature has been verified., Sıcaklık ölçümü, günlük yenidoğan bakımının hayati bir parçasıdır. Sıcaklık ölçümünde, inkübatör ve radyant ısıtıcı işlevi için beklenen değerlerden sapmaları tespit etmek amacıyla doğru ölçümlerin alınması gerekmektedir. Sıcaklığı izlemek, bebeği termonötr bir çevre bölgesinde tutmaktır. Çalışmada uygun fiyatlı kızılötesi termografi görüntüleme tekniği kullanılarak vücut yüzey sıcaklığı çeşitli klinik koşullar altında test edilmiştir. Sıcaklık dağılımları gerçek zamanlı video olarak görüntülenmesi sağlanmakta ve ortalama cilt sıcaklıklarını değerlendirmek amacıyla analiz edilmektedir. Bu çalışma, neonatologlara bir yenidoğanın cilt sıcaklığına ek güvenlik sağlamaya yardımcı olabilecek sanal bir sıcaklık algılama uygulamasının tasarımını ve uygulamasını göstermektedir. Yüzey sıcaklığı ile ilgili olarak radyant ısıtıcının içindeki farklı çevresel koşulların etkisi doğrulanmıştır.

Currently, breast cancer affects many women worldwide. In recent years, many Computer-aided diagnosis (CAD) model have been developed for early diagnosis of breast cancer. An efficient CAD model is suggested to identify mammogram images as benign versus malignant in this study. The suggested CAD model constitutes four stages which are image acgusition, segmentation, feature extraction, feature selection and classification process. Gray level run matrix (GLRM) approach is used for feature extraction, while monarch butterfly optimization (MBO) for feature selection process. Support vector machine (SVM) algorithm is preferred for classification process. The suggested model has been tested on a private mammographic dataset. The suggested model (GLRM+MBO+SVM) shows an 0.944 of accuracy for breast lesion classification. Compared with similar studies, our proposed model showed good classification results for the breast lesion classification process.

Kalp hastalığı, yaygınlığı ve yüksek ölüm oranları nedeniyle insan sağlığını tehdit etmektedir. Kalp hastalığını tahmin etmek, geleneksel yöntemler kullanarak karmaşık bir iştir. Son yıllarda, kalp hastalıklarını tahmin etmek için makine öğrenimi teknikleri kullanılmaktadır. Bu çalışma kapsamında StatLog Kalp Hastalığı veri seti üzerinde Karar Ağacı, Naive Bayes, Lojistik Regresyon, Destek Vektör Makineleri ve K-En Yakın Komşu makine öğrenme teknikleri kullanılarak kalp hastalıklarının tespitine ilişkin ilişkin karşılaştırmalı bir analiz sunulmaktadır. Veri setindeki etkin öznitelikleri seçmek için Mann Whitney U testi kullanılmıştır. Makine öğrenimi algoritmalarının sınıflandırma performansı, doğruluk, kesinlik, duyarlılık ve F1-skoru açısından değerlendirilmiştir. Destek Vektör Makineleri 96.3% doğruluk, 95.83% kesinlik, 95.83% duyarlılık ve 95.83% F1skoru ile çalışmanın en iyi tahmin oranına sahip algoritması olmuştur. Bu çalışmanın klinisyenlere kalp hastalığını erken evrede tespit etmede yardımcı olacağına inanmaktayız., Heart disease threatens human health due to its prevalence and high mortality rates. Predicting heart disease is complicated task using traditional methods. In recent years, machine learning techniques have been utilized to predict heart diseases. In this study, a comparative analysis of heart disease detection is presented using Decision Tree, Naive Bayes, Logistic Regression, Support Vector Machines and K-Nearest Neighbor machine learning techniques on the StatLog Heart Disease dataset. Mann Whitney U test is utilized to select the influential features. The classification performance of machine learning algorithms was evaluated in terms of accuracy, precision, recall and F1 score. The Support Vector Machines was the algorithm with the best prediction rate of the study, with 96.3% accuracy, 95.83% precision, 95.83% sensitivity and 95.83% F1-score. We believe that this study will help clinicians detect heart disease at an early stage.

In this study, an automated medical decision support system is presented to assist physicians with accurate and immediate brain tumor detection, segmentation, and volume estimation from MRI which is very important in the success of surgical operations and treatment of brain tumor patients. In the proposed approach, first, tumor regions on MR images are labeled by an expert radiologist. Then, an automated medical decision support system is developed to extract brain tumor boundaries and to calculate their volumes by using multimodal MR images. One advantage of this study is that it provides an automated brain tumor detection and volume estimation algorithm that does not require user interactions by determining threshold values adaptively. Another advantage is that, because of the unsupervised approach, the proposed study realized tumor detection, segmentation, and volume estimation without using very large labeled training data. A brain tumor detection and segmentation algorithm is introduced that is based on the fact that the brain consists of two symmetrical hemispheres. Two main analyses, i.e., histogram and symmetry, were performed to automatically estimate tumor volume. The threshold values used for skull stripping were computed adaptively by examining the histogram distances between T1- and T1C-weighted brain MR images. Then, a symmetry analysis between the left and right brain lobes on FLAIR images was performed for whole tumor detection. The experiments were conducted on two brain MRI datasets, i.e., TCIA and BRATS. The experimental results were compared with the labeled expert results, which is known as the gold standard, to demonstrate the efficacy of the presented method. The performance evaluation results achieved accuracy values of 89.7% and 99.0%, and a Dice similarity coefficient value of 93.0% for whole tumor detection, active core detection, and volume estimation, respectively.

Many women around the world die due to breast cancer. If breast cancer is treated in the early phase, mortality rates may significantly be reduced. Quite a number of approaches have been proposed to help in the early detection of breast cancer. A novel hybrid feature selection model is suggested in this study. This novel hybrid model aims to build an efficient feature selection method and successfully classify breast lesions. A combination of relief and binary Harris hawk optimization (BHHO) hybrid model is used for feature selection. Then, k-nearest neighbor (k-NN), support vector machine (SVM), logistic regression (LR) and naive Bayes (NB) methods are preferred for the classification task. The suggested hybrid model is tested by three different breast cancer datasets which are Wisconsin diagand mammographic breast cancer dataset (MBCD). According to the experimental results, the relief and BHHO hybrid model improves the performance of all classification algorithms in all three datasets. For WDBC, relief-BHO-SVM model shows the highest classification rates with an of accuracy of 98.77%, precision of 97.17%, recall of 99.52%, F1-score of 98.33%, specificity of 99.72% and balanced accuracy of 99.62%. For WBCD, relief-BHO-SVM model achieves of accuracy of 99.28%, precision of 98.76%, recall of 99.17%, F1-score of 98.96%, specificity of 99.56% and balanced accuracy of 99.36%. Relief-BHO-SVM model performs the best with an accuracy of 97.44%, precision of 97.41%, recall of 98.26%, F1-score of 97.84%, specificity of 97.47% and balanced accuracy of 97.86% for MBCD. Furthermore, the relief-BHO-SVM model has achieved better results than other known approaches. Compared with recent studies on breast cancer classification, the suggested hybrid method has achieved quite good results.

In this study, an automated medical decision support system is presented to assist physicians with accurate and immediate brain tumor detection, segmentation, and volume estimation from MRI which is very important in the success of surgical operations and treatment of brain tumor patients. In the proposed approach, first, tumor regions on MR images are labeled by an expert radiologist. Then, an automated medical decision support system is developed to extract brain tumor boundaries and to calculate their volumes by using multimodal MR images. One advantage of this study is that it provides an automated brain tumor detection and volume estimation algorithm that does not require user interactions by determining threshold values adaptively. Another advantage is that, because of the unsupervised approach, the proposed study realized tumor detection, segmentation, and volume estimation without using very large labeled training data. A brain tumor detection and segmentation algorithm is introduced that is based on the fact that the brain consists of two symmetrical hemispheres. Two main analyses, i.e., histogram and symmetry, were performed to automatically estimate tumor volume. The threshold values used for skull stripping were computed adaptively by examining the histogram distances between T1- and T1C-weighted brain MR images. Then, a symmetry analysis between the left and right brain lobes on FLAIR images was performed for whole tumor detection. The experiments were conducted on two brain MRI datasets, i.e., TCIA and BRATS. The experimental results were compared with the labeled expert results, which is known as the gold standard, to demonstrate the efficacy of the presented method. The performance evaluation results achieved accuracy values of 89.7% and 99.0%, and a Dice similarity coefficient value of 93.0% for whole tumor detection, active core detection, and volume estimation, respectively.

Medical Technologies Congress (TIPTEKNO) -- OCT 31-NOV 02, 2022 -- Antalya, TURKEY, This study proposes to experts a fast and highly successful algorithm for the diagnosis of ADHD disorder using EEG (Electroencephalogram) signals obtained during the Attention task, reducing their dependence on subjective evaluations. Accordingly, EEG signals obtained from 61 ADHD and 60 control participants were analyzed using nonlinear features (approximate entropy, Petrosian, and Lyapunov exponent). After feature extraction, the classification process was performed using support vector machine (SVM), and K-Nearest-Neighbor (KNN), and ensemble learning. In this study t-test based and location based feature selection methods were used. We used only features that were extracted from prefrontal and frontal regions. The highest accuracy that was reached in this study was 95.8%., Biyomedikal Klinik Muhendisligi Dernegi,Izmir Ekonomi Univ

Literatürde kullanılan tiroid nodülü risk sınıflandırma rehberleri, nodüllerin bazı iyi bilinen sonografik özelliklerine göre, hekimlerin klinik tecrübelerine dayanarak oluşturulmuşlardır. Bu özelliklere göre nodüllere tanı konması subjektif bir yöntem olup hekimin tecrübesine bağlıdır. Bu çalışmada, yapay zeka yöntemleri kullanılarak, nodüllerin ayırıcı tanısında çok çeşitli ultrason bulgularının ilişkileri incelenmiş, bu durumun üstesinden gelinmesi amaçlanmıştır. Uyarlanabilir Sinirsel Bulanık Çıkarım Sistemi'nin (ANFIS) Genetik Algoritma (GA) ile eğitimine dayalı yenilikçi bir yöntem, kötü huylu tiroid nodüllerini iyi huylu olanlardan ayırt etmek için kullanılmıştır. Önerilen yöntemden elde edilen sonuçlar yaygın olarak kullanılan ANFIS'in türev tabanlı optimize edilen algoritmaları ve Derin Sinir Ağı (DNN) yöntemi ile karşılaştırılmış, önerilen yöntemin tiroid nodüllerini sınıflandırmada daha başarılı olduğu gösterilmiştir. Ayrıca tiroid nodüllerinin sınıflandırılması için literatürde olmayan bilgisayar destekli tanı (BDT) temelli yeni bir risk sınıflandırma sistemi önerilmiştir., The thyroid nodule risk stratification guidelines used in the literature are based on certain well-known sonographic features of nodules and are still subjective since the application of these characteristics strictly depends on the reading physician. These guidelines classify nodules according to the sub-features of limited sonographic signs. This study aims to overcome these limitations by examining the relationships of a wide range of ultrasound signs in the differential diagnosis of nodules by using artificial intelligence methods. An innovative method based on training Adaptive Neuro-Fuzzy Inference Systems (ANFIS) by using Genetic Algorithm (GA) is used to differentiate malignant from benign thyroid nodules. The comparison of the results from the proposed method to the results from the commonly used derivative-based algorithms and Deep Neural Network (DNN) methods yielded that the proposed method is more successful in differentiating malignant from benign thyroid nodules. Furthermore, a novel computer aided diagnosis (CAD) based risk stratification system for the thyroid nodule's ultrasound classification that is not present in the literature is proposed.

Geçici ya da kalıcı, dahili ya da harici olarak, tedavi ve/veya onarım gibi amaçlarla vücuda dahil edilen malzemeler implant materyalleri olarak adlandırılmaktadırlar. Günümüzde artan popülasyon ve uzun yaşam süreleri göz önüne alındığında, implant materyallerin kullanım sıklığı, miktarı ve süreleri giderek artmaktadır. Bu nedenle kullanılan implantların, kullanıldıkları bölgeye uyum sağlamaları, transplantasyon sonrası herhangi bir komplikasyon oluşmasını önleyecek şekilde tasarlanmaları gerektiği sonucuna ulaşılmaktadır. Tez kapsamında polimerik ve metalik alttaşlar olmak üzere iki farklı malzemede heparin adı verilen, klinikte güncel olarak antikoagülan ilaç olarak kullanılan biyomolekül ile yüzey modifikasyon stratejilerinin geliştirilmesi üzerine çalışılmıştır. Polipropilen (PP) tez kapsamında model polimerik malzeme olarak seçilmiş ve iki basamakta plazma polimerizasyon (PlzP) destekli yüzey modifikasyonu gerçekleştirilmiştir; (i) oksijen ile aşındırma ve (ii) amince zengin fonksiyonel grup oluşturma. PlzP tekniği ile geliştirilen yüzey modifikasyon parametreleri, temas açısı ölçüm, serbest yüzey enerji hesaplamaları, yüzeyde maksimum azot miktarını elde etme, yüzey pürüzlük değerinin artırılması kriterleri kapsamında değerlendirilmiş ve optimize parametre tespit edilmiştir. Heparin immobilizasyon işlemi, saf heparin (hep) ve kovalent bağ ajanları ile destekli (hep*) çözeltiler kullanılarak 3 farklı konsantrasyon üzerinden test edilmiştir. Yüzeye tutunumu sağlanan heparin miktarı Toluidine Mavi (TB) boya ile tespit edilmiş ve maksimum heparin tutunumunun sağlandığı konsantrasyon, optimize parametre olarak belirlenmiştir. Kovalent bağ ajanları ile destekli heparin çözeltisinin iki basamakta modifiye edilen PP yüzeyler ile kovalent bağ oluşturduğu, FTIR-ATR analizinde saptanan amid I ve II pikleri ile kanıtlanmıştır. Heparin immobilizasyonu protein tutunum miktarını referans yüzeylere göre oldukça azaltmış, hemokompatibilite testleri kapsamında (platelet tutunum, Implantable materials are tools that are used in the human body for temporary or permanent, internal or external, treatment or repair, and so on. Nowadays, due to the growing population and long lifetime, the rate, quantity, and duration of these materials are also increasing. Therefore, implantable materials should be biocompatible with the environment in their usage area and be designed to prevent any complications after the transplant. Within the scope of this thesis, two different materials were used as polymeric and metallic substrates for surface modification with an anticoagulation drug that is used in clinics, heparin. Polypropylene (PP) was selected as a model polymeric substrate and plasma-assisted surface modification was applied in two steps: (i) oxygen etching and (ii) amine-rich functional group creation. Surface modification parameters that were held with the plasma polymerization (PlzP) technique were analyzed with contact angle measurements, surface free energy calculations, obtaining the maximum nitrogen amount on the surfaces and increasing the surface roughness values, etc., and the parameters were optimized accordingly. Immobilization of heparin applied as pure heparin (hep) and covalent bond agent assisted (hep*) with 3 different concentration values. The amount of immobilized heparin detected with toluidine blue (TB) dye and the parameter that produces the maximum amount of immobilized heparin were selected as optimal. Covalent bond formation between hep* and PP substrates was proven with amide I and II peaks, which were detected with FTIR-ATR analysis. Compared to the bare PP substrates, heparin immobilization provided a lower amount of adhered protein and much more efficient hemocompatibility testing (platelet adhesion, hemolysis, and kinetic blood coagulation rate) results. Also, the antiadhesive characteristics of heparin-immobilized PP surfaces were tested against both gram-positive and gram-negative bacteria strains, and the antibacteria

Meme kanseri dünya genelinde kadınlar arasında en sık görülen kanser türüdür. Meme kanseri erken evrede teşhis edilirse, tedavi edilmesi mümkündür. Bu çalışma meme kanserinin tanısı için geliştirilmiş makine öğrenme algoritmalarına dayalı yeni bir sınıflandırma sistemi önermektedir. Geliştirilmiş makine öğrenme algoritmaları oluşturmak amacıyla öznitelik seçim ve hiperparametre optimizasyon yöntemleri kullanılmıştır. Makine öğrenme algoritması olarak sırasıyla Karar Ağacı, Naive Bayes, Destek Vektör Makinesi, K-En Yakın Komşu ve Topluluk Öğrenme yöntemleri kullanılmıştır. Tüm deneyler Wisconsin Meme Kanseri Veri (WBCD) seti ve Mamografi Meme Kanseri Veri Seti (MBCD) olmak üzere iki farklı meme kanseri veri seti üzerinde test edilmiştir. Veri setlerinin en ayırt edici özniteliklerini belirlemek amacıyla sırasıyla Relief, En Küçük Mutlak Daralma ve Seçme Operatörü ((Least Absolute Deviation and Least Absolute Shrinkage and Selection Operator-LASSO) ve Ardışık İleri Yönde Seçim yöntemleri kullanılmıştır. Makine öğrenme algoritmalarındaki en uygun hiperparametreleri bulmak için Bayes optimizasyon (BO) yöntemi kullanılmıştır. Çalışma kapsamında en iyi sınıflandırma oranını elde etmek amacıyla farklı deneyler yapılmıştır. Önerilen öznitelik seçim-Bayes optimizasyon hibrit yöntemleri çalışmada kullanılan makine öğrenme algoritmalarının sınıflandırma oranlarını artırmıştır. Yapılan deneyler sonucunda, LASSO-BO-DVM yöntemi her iki meme kanseri veri setinde de en yüksek doğruluk, kesinlik, duyarlılık ve F1-skorunu göstermiştir (WBCD için %98,95, %97,17, %100 ve %98,56; MBCD için %97,95, %98,28, %98,28 ve %98,28)., Breast cancer is the most common cancer type among women worldwide. If breast cancer is detected at an early stage, it can be cured. This study proposes a novel classification model based improved machine learning algorithms for diagnosis of breast cancer. Feature selection and hyperparameter optimization methods are used to build improved the machine learning algorithms. Decision Tree, Naive Bayes, Support Vector Machine, K-Nearest Neighbor and Essemble Learning methods are used as machine learning algorithms, respectively. All experiments are tested on two different datasets, Wisconsin Breast Cancer Dataset (WBCD) and Mammographic Breast Cancer Dataset (MBCD). Relief, Least Absolute Deviation and Least Absolute Shrinkage and Selection Operator (LASSO) and Sequential Forward Selection methods are used to determine the most distinctive features of the datasets, respectively. Bayesian optimization (BO) method is used to find optimal hyperparameters in machine learning algorithms. Within the scope of this study, different experiments are conducted in order to obtain the best classification rate. The proposed feature selection-Bayes optimization hybrid methods have increased the classification rates of the machine learning algorithms used in the study. As a result of the experiments, LASSO-BO-SVM has showed the highest accuracy, precision, recall and F1-score in both datasets (%98,95, %97,17, %100, %98,56 for WBCD; %97.95, %98,28, %98,28, %98,28 for MBCD).

Patolojik duruma bağlı olarak normal konuşma akışının bozulması, ses bozukluğu olarak bilinir. Bu nedenle, mevcut herhangi bir bozukluk, konuşma üretim sisteminin işleyişini bozar ve dolayısıyla bozuk bir ses üretir. Bazı laringeal patolojiler hayatı tehdit eder, bu nedenle ses bozukluğunun erken tespiti önemlidir. Patolojik seslerin tespitinde bir karar destek sisteminin geliştirilmesi hayati önem taşımaktadır. Patolojik seslerin belirlenmesi amacıyla seslerden çıkarılan özniteliklerin değerlendirilmesinde istatistiksel yöntemlerin grup bazında bir sonuç vermesi nedeniyle bireysel düzeyde bir cevap elde edilebilmesi amacıyla son yıllarda makine öğrenme yöntemleri araştırmacılar tarafından ilgi çekici bir konu olmuştur. Bununla birlikte makine öğrenmesinin özniteliklerin manuel çıkarılmasına ihtiyaç duyması nedeniyle optimal özniteliklerin otomatik olarak çıkarılabildiği derin öğrenme teknikleri araştırmacıların güncel araştırma konuları arasına girmiştir. Ancak henüz patolojik ses bozukluklarının tespiti alanında derin öğrenme tekniklerinin kullanımı ile ilgili az sayıda araştırma çalışması bulunmaktadır. Bu tez çalışmasında, patolojik seslerin belirlenmesi amacıyla derin öğrenme yöntemleri kullanılmıştır. Çalışmada Saarbruecken Ses Veritabanından vokal kordlardaki yapısal değişikliklerin neden olduğu organik disfoniye sebep olan patolojilere sahip hastaların ses kayıtları seçilmiştir. Bu patolojiler arasında larenjit, lökoplazi, Reinke ödemi, rekürren laringeal sinir felci, vokal kord karsinomu ve vokal kord polibi bulunmaktadır. Her bir bireyin nötr perdesinde sürekli sesli /a/ sesi kayıtları seçilmiştir. 380'i sağlıklı ve 380'i patolojik olmak üzere 760 ses kaydı kullanılmıştır. Veriler, sırasıyla %75 ve %25 örnek içeren eğitim seti ve test seti olarak ayrılmıştır. Ses sinyallerine öncelikle dalgacık gürültü giderme işlemi uygulanmıştır. Daha sonrasında ses sinyallerinin spektrogram görüntüleri alınarak dört faklı Evrişimsel Sinir Ağı (ESA) mimarisine girdi olar, The disruption of normal speech flow due to pathological conditions is known as a voice disorder. Therefore, any existing disorder disrupts the speech production system's functioning and produces a distorted voice. Since some laryngeal pathologies are life-threatening, the early detection of voice disorders is important. For this purpose, there is a need to develop a decision support system in the detection of pathological voices. In recent years, machine learning methods have become an interesting research topic to determine pathological voices in order to obtain an individual-level answer, since statistical methods give a group-based result in the evaluation of features extracted from voices. However, since machine learning requires manual extraction of features, deep learning techniques, in which optimal features can be extracted automatically, have become one of the current research topics. However, there are only few research studies on the use of deep learning techniques in the detection of pathological voice disorders. In this thesis study, deep learning methods were used to identify pathological voices. The voice recordings of patients with pathologies causing organic dysphonia due to structural changes in the vocal cords were selected from the Saarbruecken Voice Database. These pathologies included laryngitis, leukoplakia, Reinke's edema, recurrent laryngeal nerve paralysis, vocal cord carcinoma, and vocal cord polyps. The sustained vowel /a/ at the neutral pitch of each individual was selected. The sample included a total of 760 recordings, of which 380 belonged to healthy voices and 380 belonged to pathological voices. The data were divided into training and test sets containing 75% and 25% of the samples, respectively. In the analysis of the samples, first, wavelet noise denoising was applied to the voice signals. Then, the spectrogram images of the voice signals were taken and utilized as inputs in four different Convolutional Neural Network (CNN) archi

Image analysis of cell biology and tissue engineering is time-consuming and requires personal expertise. However, evalu - ation of the results may be subjective. Therefore, computer-based learning and detection applications have been rapidly developed in recent years. In this study, Confocal Laser Scanning Microscope (CLSM) images of the viable pre-osteoblastic mouse MC3T3-E1 cells in 3D bioprinted tissue scaffolds, captured from a bone tissue regeneration study, were analyzed by using image processing techniques. The aim of this study is to develop a reliable and fast algorithm for the automated analysis of live/dead assay CLSM images. Percentages of live and dead cell areas in the scaffolds were determined, and then, total cell viabilities were calculated. Furthermore, manual measurements of four different analysts were obtained to evaluate subjectivity in the analysis. The measurement variations of analysts, also known as the coefficient of variation, were determined from 13.18% to 98.34% for live cell images and from 9.75% to 126.02% for dead cell images. Therefore, an automated algorithm was developed to overcome this subjectivity. The other aim of this study is to determine the depth profile of viable cells in 3D tissue scaffolds. Consequently, cross-sectional image sets of three different types of tissue scaf - folds were analyzed.

Ağrı, haz verici olmayan kompleks ve çok boyutlu olmak üzere kişilerin tecrübe ettiği his olarak tanımlanır. Ağrının objektif olarak sınıflandırılması, hastanelerde mönitörize olmuş hastalar için, kendini ifade edemeyen bireyler için ve ameliyat sonrası ya da esnasında anesteziye maruz kalmış kişiler için büyük önem taşımaktadır. Ayrıca ağrının subjektif olarak derecelendirilmesi kişiden kişiye yanıltıcı bir cevaba sebep olabilmektedir ve kişinin yanlış tedavi protkolüne yönlendirilmesi ile sonuçlanabilmektedir. Ağrının yapay zeka modelleri ile derecelendirilmesi sayesinde doğru tedaviye yönelim ve kendini ifade edemeyen bireylerin ağrı şiddetlerini derecelendirebilmesi hakkında bilgiye ulaşmak mümkün olabilecektir. Yapılan tez çalışmasında, ağrı uyarını verilerek bireylerden alınan EEG verileri bir boyutlu evrişimsel sinir ağı modeli (1 Dimensional Convolutional Neural Networks, 1D CNN) oluşturularak yüksek ve düşük ağrı olarak sınıflandırılmıştır. Modelde kullanılacak öznitelik seçimi için alınan EEG verileri öncesinde gürültülerden temizlenmiştir. Sonraki aşamada temizlenen verilere T-test uygulanarak fark oluşan bölümler belirlenmiştir. Belirlenen bölümlerden zaman-frekans cevabında öznitelikler çıkarılarak hazırlanan modelde kullanılmıştır. Tasarlanan model karşılaştırılma amaçlı derin öğrenme modellerinden olan özyinemeli sinir ağları (Reccurrent Neural Network, RNN) modeli ile karşılaştırılmıştır. Tez çalışmasının sonucunda alınan bilgiler dahilinde tasarlanan 1 boyutlu evrişimsel sinir ağları modelinin ortalama %92 oranında doğru sınıflandırma yaptığı gözlemlenmiştir. Elde edilen bulgular dahilinde ağrı hissinin objektif derecelendirilmesi için tasarlanan modelin iyi bir belirteç olabileciğini göstermektedir., Pain is defined as a non-pleasurable, complex and multidimensional sensation experienced by people. Objective classification of pain is of great importance for patients who are monitored in hospitals, for individuals who cannot express themselves, and for people who have been exposed to anesthesia during or after surgery. In addition, subjective grading of pain can cause a misleading response from person to person and may result in the person being directed to the wrong treatment protocol. Thanks to the grading of pain with artificial intelligence models, it will be possible to reach information about orientation to the right treatment and grading the severity of pain of individuals who cannot express themselves. In the thesis study, EEG data obtained from individuals by giving pain stimulus was classified as high and low pain by creating a one-dimensional convolutional neural network model (1 Dimensional Convolutional Neural Networks, 1D CNN). The EEG data obtained for the feature selection to be used in the model were cleared of noise beforehand. In the next step, T-test was applied to the cleaned data and the parts that made a difference were determined. The features in the time-frequency response from the determined sections were extracted and used in the prepared model. The designed model was compared with the Reccurrent Neural Network (RNN) model, which is one of the deep learning models for comparison purposes. As a result of the thesis study, it was observed that the 1D convolutional neural network model, designed within the scope of the information received, made an average of 92% correct classification. The findings show that the model designed for objective grading of pain sensation can be a good marker.

Medical Technologies Congress (TIPTEKNO) -- OCT 31-NOV 02, 2022 -- Antalya, TURKEY, The infant mortality rate is an important indicator of maternal and infant health, as well as global health status in general. In 2018, annual infant deaths were announced at 4.0 million. The main causes must be pointed out for decreasing infant deaths and unexplained death situations (SIDS). Infants' thermoregulation systems malfunctioning or hypothermia-hyperthermia situations are proposed as the primary or ultimate causes of SIDS. On these foundations, before generating a model, developed thermoregulation system models were examined. Using infant-specific parameters and more precise ambient conditions, the model depth was increased. To investigate the thermal distribution of the infant body, the infant body was divided into 7 compartments, which were then sliced into layers. The thermoregulation system of infants was simulated using thermodynamic equations and generated equations. The model's accuracy was tested by comparing its findings to those from experiments and data from the literature. The results unmistakably demonstrated that the model's data and those gathered from the real system and the literature are in good agreement., Financial support for this study by the Scientific and Technological Research Council of Turkey (TUBITAK) with B2210-D National Industrial MSc/MA Scholarship Program under grant no. 1649B022100041 is gratefully acknowledged. I'd like to express my gratitude to Prof. Dr. Osman Ero.gul, my advisor/consultant. This dissertation would not have been possible without his guidance and persistent assistance. I'd like to thank OKUMAN Medikal Sistemler A.S.. for their assistance and support in this study. Last but not least, I'd like to express my heartfelt appreciation for my mother's unending support., Biyomedikal Klinik Muhendisligi Dernegi,Izmir Ekonomi Univ, Scientific and Technological Research Council of Turkey (TUBITAK) [1649B022100041]

Temporal lobe epilepsy, a neurological disease that causes seizures as a result of excessive neural activities in the brain, is the most common type of focal seizure, accounting for 30-35% of all epilepsies. Detection of epilepsy and localization of epileptic focus are essential for treatment planning and epilepsy surgery. Currently, epileptic focus is decided by expert physician by examining the EEG records and determining EEG channel where epileptic patterns begins and continues intensely during seizure. Examination of long EEG recordings is very time-consuming process, requires attention and decision can vary depending on physician. In this study, to assist physicians in detecting epileptic focus side from EEG recordings, a novel deep learning-based computer-aided diagnosis system is presented. In the proposed framework, ictal epochs are detected using long short-term memory network fed with EEG subband features obtained by discrete wavelet transform, and then, epileptic focus identification is realized by using asymmetry score. This algorithm was tested on EEG database obtained from the Ankara University hospital. Experimental results showed ictal and interictal epochs were classified with accuracy of 86.84%, sensitivity of 86.96% and specificity of 89.68% on Ankara University hospital dataset, and 96.67% success rate was obtained on Bonn EEG dataset. In addition, epileptic focus was identified with accuracy of 96.10%, sensitivity of 100% and specificity of 93.80% by using the proposed deep learning-based algorithm and university hospital dataset. These results showed that proposed method can be used properly in clinical applications, epilepsy treatment and surgical planning as a medical decision support system.

Sternokostal eklemler, toraks bölgesinde ilk yedi kaburga ile sternum arasında yer alan eklemlerdir. Göğüs duvarında meydana gelen tümörler, enfeksiyonlar ve travmaların tedavisi sternumun ve ona bağlı sternokostal eklemlerin cerrahi olarak çıkartılması ile sonuçlanabilmektedir. Hastanın yaşamsal fonksiyonlarının devam etmesini sağlamak için çıkarılan hasarlı bölgenin yerine geçebilecek kişinin dokusu ile uyumlu malzemeden üretilmiş sternokostal implantlar kullanılabilmektedir. Pazardaki standart implantların yanı sıra kişiye özel tasarlanan implantlar ameliyatların risklerini azaltmanın yanı sıra cerrahların işini de kolaylaştırmaktadır. Bilgisayar ortamında tasarlanan kişiye özel bu implantlar eklemeli/katmanlı üretim (Additive Manufacturing) olarak kabul edilen 3 boyutlu yazıcı teknolojileri sayesinde somut hale dönüştürülebilmektedir. Bu çalışmada 3 boyutlu üretim teknolojilerinden toz esaslı eklemeli imalat yöntemi olan SLS (Seçici Lazer Sinterleme) ile kişiye özel sternokostal eklem implantının tasarım ve üretim aşamaları incelenmiştir. METÜM (Medikal Tasarım ve Üretim Merkezi, Sağlık Bilimleri Üniversitesi Gülhane Yerleşkesi, Ankara) tarafından oluşturulan ve hastaya implantasyonu gerçekleştirilen Ti-6Al-4v malzemeden üretilmiş bir sternokostal implantın analiz programında sonlu elemanlar analizi yapılmıştır. Hastaya ait tomografi görüntüleri alınarak 3 boyutlu modelleme programında göğüs kafesi modeli oluşturulmuş ve dijital ortamda hazırlanan implant modeli ile birlikte analiz programında statik analizleri gerçekleştirilmiştir. Analizde belirli sınırlı şartları oluşturulmuş ve tüm modellerde aynı sınır şartlarında implant üzerindeki stres, gerinim ve yer değiştirme (deformasyon) değerleri incelenmiştir. Elde edilen sonuçlara göre göğüs kafesine entegre edilmiş implant modeline bir CPR kuvveti uygulandığında implantın bu kuvvete dayanabildiği, stres sonuçlarının titanyumun akma dayanımının altında kaldığı dolayısıyla implantta kalıcı şekil değişikliğine sebe, The sternocostal joints are the joints located in the thoracic region between the first seven ribs and the sternum. Some of the tumors, infections, and traumas occurring in the chest wall result in reconstruction of the sternum and its associated sternocostal joints. In order to enhance the patient's quality of life, sternocostal implants made of anatomical and material compatible with the tissue of the person to be used to replace the removed damaged area. Unlike the standard implants in the market, the production of implants specific to the person's condition not only reduces the risks of the surgeries but also facilitates the job of the surgeons. These implants, which are designed in computer aided technoloiges, can be turned into tangible 3-D models thanks to 3-D printer technologies, which are accepted as additive manufacturing. In this study, the production and design stages of a custom sternocostal joint implant with SLS (Selective Laser Sintering), which is a powder-based additive manufacturing method of 3-D manufacturing technologies, was examined and a new implant design was proposed. In the analysis program of a sternocostal implant made of Ti-6Al-4V material created by MDMC (Medical Design and Manufacturing Center, University of Health Sciences, Gülhane, Ankara) and implanted to the patient, finite element analysis was performed and the loads and stress values on the implant were measured. In the analysis, specific bounded conditions were created and stress, strain and deformation values on the implant were examined under the same boundary conditions in all models. According to the results, it was observed that when a CPR force is applied to the implant model integrated into the rib cage, the implant can withstand this force, and the stress results remain below the yield strength of titanium, so it does not cause permanent deformation of the implant. Some sternocostal implants designed to date in the world and in our country have been examined within the

The inferior hemocompatibility or antibacterial properties of blood-contacting materials and devices are restraining factors that hinder their successful clinical utilization. To highlight these, a plasma-enhanced modification strategy is favored for surface tailoring of an extensively used biomaterial, polypropylene (PP). The surface activation of the PPs is achieved by oxygen plasma etching and subsequent surface functionalization through amine-rich precursor mediated coating by plasma glow discharge. After optimum plasma processing parameters are decided, heparin (anticoagulant and antithrombic drug) is either attached or covalently conjugated on the PPs' surfaces. The aminated films produced at 75 W plasma power with 15 min exposure time are highly hydrophilic (34.72 +/- 5.92 degrees) and surface active (65.91 mJ m(-2)), facilitating high capacity heparin immobilization (approximate to 440 mu g cm(-2)) by covalent linkage. The kinetic-blood coagulation rate and protein adhesion amount on the plasma-mediated heparinized PPs are decreased about tenfold and 15-fold, and platelet adhesion is markedly lowered. In addition, heparinized-PP surfaces comprise superior antibacterial activity against gram-positive/-negative bacteria conveyed particularly by contact-killing (99%). The heparin-coating did not cause cytotoxicity on fibroblast cells, instead enhanced their proliferation, as shown by the (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide) assay. Overall, this simple methodology is highly proficient in becoming a universal strategy for developing dual-functionalized blood-contacting materials.

Yenidoğan ölüm hızı, anne ve bebek sağlığının yanı sıra genel olarak küresel sağlık durumunun önemli bir göstergesidir. Küresel bazda yayımlanan birçok raporda yenidoğan ölüm oranı (ingilizce çeviri, Infant Mortality Rate), sağlık alanında gerçekleşen teknolojik ve yöntemsel gelişmelerin takibi için kullanılmaktadır. Dünya Sağlık Örgütü'ne ait rapora göre, 2020'de yaşamlarının ilk yılında 2.4 milyon yenidoğan çeşitli sebeplerden dolayı hayatını kaybetmiştir. Bu sayı, günde yaklaşık 6.700 yenidoğan ölümüne işaret etmektedir. Her yıl tahminen 15 milyon yenidoğan, prematüre (37. gebelik haftasından önce) olarak doğmaktadır ve bu rakam gün geçtikçe artmaktadır. Yenidoğan ölümü, erken doğum ve hastalıklı yenidoğan doğumu oranlarının azaltılması ve bu konular hakkında geniş, detaylı, empirik bilgiler edinilmesi büyük önem arz etmektedir. Bu nedenle yenidoğanlar üzerinde gerçekleşen klinik araştırmaların genişletilmesi, tedavi yöntemleri ve bakım prosedürlerinin iyileştirilmesi gerekmektedir. Yenidoğan mortalitesi, morbiditesi ve nedenleri ile alakalı gerçekleştirilen literatür taramasında, görülen komplikasyonların birbirini tetikler nitelikte olduğu görülmüştür. Olgunlaşmamış, diğer bir tabirle fonksiyonelliğini tam olarak kazanmamış, birçok fizyolojik sisteme sahip yenidoğan, yaşamının ilk yıllarında birden fazla stres kaynağına maruz kalmaktadır. Bu stres kaynaklarından en yüksek insidans oranına sahip kaynak, termal stres olarak karşımıza çıkmaktadır. Buna ek olarak ani yenidoğan ölüm sendromunun en büyük kaynağı ısıl stres olarak gösterilmektedir. Bu çalışmada yenidoğan termoregülasyon sistemi, yenidoğan üzerinde gerçekleşen ısı transferleri ve mekanizmaları, karşılaşılan termal stres kaynaklı vakalar, yenidoğan yoğun bakım ya da bakım odalarında kullanılan ısıl destek üniteleri, yenidoğanın ısı kaybetmemesi için alınması gereken önlemler, yöntemler incelenmiştir. Gerçekleştirilen çalışmada yenidoğan termoregülasyon sistemi için biyolojik bir sistem modeli oluşturulm, The infant (or neonatal) mortality rate is an important indicator of maternal and infant health, as well as global health in general. In many reports published on the global basis, the infant mortality rate is used to monitor technological and methodological developments in the field of health. According to a report by the World Health Organization, 2.4 million newborns died in their first year of life in 2020. This number represents approximately 6,700 neonatal deaths per day. An estimated 15 million newborns are born prematurely (before 37 weeks of gestation) each year, and this number is increasing day by day. It is of great importance to reduce the rates of neonatal death, preterm birth, and birth with disease(s) and to obtain extensive, detailed, and empirical information on these issues. Therefore, it is necessary to expand clinical studies on newborns, improve treatment methods and care procedures. In the literature review on neonatal mortality, morbidity, and their causes, it was seen that the complications observed were triggering each other. Immature, or in other words, not fully functioning, physiological systems of infants are exposed to more than one stress source in the first years of their lives. The source with the highest incidence rate among these stress sources is thermal stress. In addition, thermal stress is considered as the biggest source of sudden infant death syndrome by the official sources. In this study, newborn thermoregulation system, heat transfers and mechanisms on the newborn, cases caused by thermal stress, thermal support units used in neonatal intensive care or care rooms, precautions and methods to be taken to prevent the newborn from losing heat were examined. In this study, before creating a biological system model for the neonatal thermoregulation system, neonatal and adult human thermoregulation system models in the literature were reviewed. In the developed biological model, many of the assumptions or generalizations used in

Sleep apnea is defined as the cessation of breathing for at least ten seconds and known as a common sleep disorder. Obstructive sleep apnea (OSA) is the most common type of sleep apnea that occurs due to obstruction in the airway. Besides the detection of sleep apnea with the help of developed algorithms, early prediction of this syndrome is important in order to prevent serious health problems and life-threatening situations. With the prediction models working with high accuracy, it will be possible to prevent the possible risks without experiencing them by stimulating the OSA patients before the syndrome occurs and waking them up from sleep. In this thesis, models that predict apnea using electrocardiographic signals of patients diagnosed with obstructive slep apnea by using Convolutional Neural Networks (CNN) are presented. The first model is the training from scratch with pre-trained architectures. The second model is the study of using the transfer learning method with pre-trained architectures. The third model is the study in which the new results observed with the changes made on the architecture that performed the best prediction performance in the first two models. In the last model, the results are presented when the features obtained from the deep learning architecture proposed in the third model are classified by Support Vector Machines, Random Subspace k-Nearest Neighborhood and Random Subspace Discriminant Analysis methods instead of the architecture's own classifier. The high accuracy findings observed at the end of the study show that the proposed model can be used as a good indicator for the prediction of OSA syndrome., Nefes alışverişinin en az on saniye boyunca durması olarak tanımlanan uyku apnesi, günümüzde sık karşılaşılan bir uyku hastalığı olarak bilinmektedir. Obstrüktif uyku apnesi (OUA), solunum yolunda tıkanmaya bağlı gerçekleşen en yaygın uyku hastalıklarından biridir. Bu sendromun geliştirilen modeller yardımıyla otomatik olarak tespit edilebilmesinin yanında ön görülebilmesi de ciddi seviyelerde sağlık problemlerini ve hayati tehlikeyle karşı karşıya kalma durumunu önlemek açısından önemlidir. Yüksek doğrulukla çalışan tahmin modellerinin geliştirilmesi ile OUA rahatsızlığı yaşayan kişilerin sendrom anı gelmeden uyarılması ve uykudan uyandırılması ile olası risklerin yaşanmadan önlenmesi mümkün olabilecektir. Bu tez çalışmasında, derin öğrenme yöntemlerinden biri olan Evrişimsel Sinir Ağları (Convolutional Neural Networks, CNN) ile OUA tanısı konmuş hastalara ait elektrokardiyografi sinyalleri kullanılarak apne tahmini yapan modeller sunulmuştur. Bu modellerden birincisi, önceden eğitilmiş mimariler ile yapılan sıfırdan eğitim çalışmasıdır. İkinci model, önceden eğitilmiş mimariler ile transfer öğrenme yöntemi kullanılarak yapılan çalışmadır. Üçüncü model, ilk iki modelde gözlemlenen bulgulara bağlı olarak önerilmiş yeni bir modeldir. Son modelde ise üçüncü modelde önerilen derin öğrenme mimarisinden elde edilen özniteliklerin, mimarinin kendi sınıflandırıcısı yerine Destek Vektör Makineleri, Rastgele Alt Uzay k-En yakın Komşuluk ve Rastgele Alt Uzay Diskriminant Analizi yöntemleri ile sınıflandırıldığı durumda gözlemlenen sonuçlar sunulmuştur. Çalışmanın sonucunda gözlemlenen yüksek doğruluktaki bulgular, önerilen modellerin OUA tahmininde iyi bir belirteç olarak kullanılabileceğini göstermektedir.

The thyroid nodule risk stratification guidelines used in the literature are based on certain well-known sonographic features of nodules and are still subjective since the application of these characteristics strictly depends on the reading physician. These guidelines classify nodules according to the sub-features of limited sonographic signs. This study aims to overcome these limitations by examining the relationships of a wide range of ultrasound (US) signs in the differential diagnosis of nodules by using artificial intelligence methods. An innovative method based on training Adaptive-Network Based Fuzzy Inference Systems (ANFIS) by using Genetic Algorithm (GA) is used to differentiate malignant from benign thyroid nodules. The comparison of the results from the proposed method to the results from the commonly used derivative-based algorithms and Deep Neural Network (DNN) methods yielded that the proposed method is more successful in differentiating malignant from benign thyroid nodules. Furthermore, a novel computer aided diagnosis (CAD) based risk stratification system for the thyroid nodule's US classification that is not present in the literature is proposed.

Medical Technologies Congress (TIPTEKNO) -- OCT 31-NOV 02, 2022 -- Antalya, TURKEY, This study proposes to experts a fast and highly successful algorithm for the diagnosis of ADHD disorder using EEG (Electroencephalogram) signals obtained during the Attention task, reducing their dependence on subjective evaluations. Accordingly, EEG signals obtained from 61 ADHD and 60 control participants were analyzed using nonlinear features (approximate entropy, Petrosian, and Lyapunov exponent). After feature extraction, the classification process was performed using support vector machine (SVM), and K-Nearest-Neighbor (KNN), and ensemble learning. In this study t-test based and location based feature selection methods were used. We used only features that were extracted from prefrontal and frontal regions. The highest accuracy that was reached in this study was 95.8%., Biyomedikal Klinik Muhendisligi Dernegi,Izmir Ekonomi Univ

Image analysis of cell biology and tissue engineering is time-consuming and requires personal expertise. However, evalu - ation of the results may be subjective. Therefore, computer-based learning and detection applications have been rapidly developed in recent years. In this study, Confocal Laser Scanning Microscope (CLSM) images of the viable pre-osteoblastic mouse MC3T3-E1 cells in 3D bioprinted tissue scaffolds, captured from a bone tissue regeneration study, were analyzed by using image processing techniques. The aim of this study is to develop a reliable and fast algorithm for the automated analysis of live/dead assay CLSM images. Percentages of live and dead cell areas in the scaffolds were determined, and then, total cell viabilities were calculated. Furthermore, manual measurements of four different analysts were obtained to evaluate subjectivity in the analysis. The measurement variations of analysts, also known as the coefficient of variation, were determined from 13.18% to 98.34% for live cell images and from 9.75% to 126.02% for dead cell images. Therefore, an automated algorithm was developed to overcome this subjectivity. The other aim of this study is to determine the depth profile of viable cells in 3D tissue scaffolds. Consequently, cross-sectional image sets of three different types of tissue scaf - folds were analyzed.

Medical Technologies Congress (TIPTEKNO) -- OCT 31-NOV 02, 2022 -- Antalya, TURKEY, The infant mortality rate is an important indicator of maternal and infant health, as well as global health status in general. In 2018, annual infant deaths were announced at 4.0 million. The main causes must be pointed out for decreasing infant deaths and unexplained death situations (SIDS). Infants' thermoregulation systems malfunctioning or hypothermia-hyperthermia situations are proposed as the primary or ultimate causes of SIDS. On these foundations, before generating a model, developed thermoregulation system models were examined. Using infant-specific parameters and more precise ambient conditions, the model depth was increased. To investigate the thermal distribution of the infant body, the infant body was divided into 7 compartments, which were then sliced into layers. The thermoregulation system of infants was simulated using thermodynamic equations and generated equations. The model's accuracy was tested by comparing its findings to those from experiments and data from the literature. The results unmistakably demonstrated that the model's data and those gathered from the real system and the literature are in good agreement., Financial support for this study by the Scientific and Technological Research Council of Turkey (TUBITAK) with B2210-D National Industrial MSc/MA Scholarship Program under grant no. 1649B022100041 is gratefully acknowledged. I'd like to express my gratitude to Prof. Dr. Osman Ero.gul, my advisor/consultant. This dissertation would not have been possible without his guidance and persistent assistance. I'd like to thank OKUMAN Medikal Sistemler A.S.. for their assistance and support in this study. Last but not least, I'd like to express my heartfelt appreciation for my mother's unending support., Biyomedikal Klinik Muhendisligi Dernegi,Izmir Ekonomi Univ, Scientific and Technological Research Council of Turkey (TUBITAK) [1649B022100041]