Profile URL: https://hdl.handle.net/20.500.13091/11763
Job Title:Arş. Gör. Uzm.
Email Address:ekurnaz@ktun.edu.tr
Main Affiliation:02.04. Department of Electrical and Electronics Engineering
Status: Current Staff
ORCID:
0000-0002-6671-348X
0000-0002-6671-348XScopus ID:
57205200677
57205200677YÖK Akademik: 758D00AD073EB286
Google Scholar:
mgdyRPYAAAAJ
mgdyRPYAAAAJWeb of Science ID:
ABG-7613-2020
ABG-7613-20209 results
Scholarly Output Search Results
Now showing 1 - 9 of 9
Article Citation - WoS: 1Citation - Scopus: 1A Novel Deep Learning Model for Pancreas Segmentation: Pascal U-Net(Asoc Espanola Inteligencia Artificial, 2024-05-17) Kurnaz, Ender; Ceylan, Rahime; Bozkurt, Mustafa Alper; Cebeci, Hakan; Koplay, MustafaA robust and reliable automated organ segmentation from abdomen images is a crucial problem in both quantitative imaging analysis and computer-aided diagnosis. In particular, automatic pancreas segmentation from abdomen CT images is the most challenging task based on two main aspects (1) high variability in anatomy (like as shape, size, etc.) and location across different patients and (2) low contrast with neighbouring tissues. Due to these reasons, the achievement of high accuracies in pancreas segmentation is a hard image segmentation problem. In this paper, we propose a novel deep learning model which is a convolutional neural network-based model called Pascal U-Net for pancreas segmentation. The performance of the proposed model is evaluated on The Cancer Imaging Archive (TCIA) Pancreas CT database and abdomen CT dataset which is taken from Selcuk University Medicine Faculty Radiology Department. During the experimental studies, the k-fold cross-validation method is used. Furthermore, the results of the proposed model are compared with the results of traditional U-Net. If results obtained by Pascal U-Net and traditional U-Net for different batch sizes and fold number is compared, it can be seen that experiments on both datasets validate the effectiveness of the Pascal U-Net model for pancreas segmentation.Conference Object Citation - WoS: 14Citation - Scopus: 21Design and Simulation of the Hierarchical Tree Topology Based Wireless Drone Networks(IEEE, 2018-09-01) Çeltek, Seyit Alperen; Durdu, Akif; Kurnaz, EnderIn drone applications, the drone could send data using telemetry devices or radio frequency module which has a limited range. So, there is no interaction between user and drone after a certain range. In this study, a hierarchical tree topology based wireless drone network is designed and presented to overcome range challenge. Proposed network consist of three main parts; Control Center (CC), Master Drone (MD) and Slave Drones (SDs). The CC as a network manager communicates with just MD via telemetry devices. SDs are explorer drones for the search and rescue application. The data transfer between CC and SDs is provided by MD which is explorer like SDs. This paper clearly shows that the enhancement of the communication range is possible with such this approach. Also the designed drone networks are simulated using V-REP (Virtual Robot Experimentation Platform). According to the simulation results, the proposed drone network system operates quickly, and finds the target in 5 minutes, which classical system not find in 10 minutes. The proposed model clearly shows that an application using a drone is completed in a shorter time with the drone swarm well organized.Article Image Processing Based Task Allocation for Autonomous Multi Rotor Unmanned Aerial Vehicles(SCIENCE & INFORMATION SAI ORGANIZATION LTD, 2018) Durdu, Akif; Ergene, Mehmet Celalettin; Demircan, Onur; Uğuz, Hasan; Mahmutoglu, Mustafa; Kurnaz, EnderNowadays studies based on unmanned aerial vehicles draws attention. Especially image processing based tasks are quite important. In this study, several tasks were performed based on the autonomous flight, image processing and load drop capabilities of the Unmanned Aerial Vehicle (UAV). Two main tasks were tested with an autonomous UAV, and the performance of the whole system was measured according to the duration and the methods of the image processing. In the first mission, the UAV flew over a 4x4 sized color matrix. 16 tiles of the matrix had three main colors, and the pattern was changed three times. The UAV was sent to the matrix, recognized 48 colors of the matrix and returned to the launch position autonomously. The second mission was to test load drop and image processing abilities of the UAV. In this mission, the UAV flew over the matrix, read the pattern and went to the parachute drop area. After that, the load was dropped according to the recognized pattern by the UAV and then came back to the launch position.Conference Object Addressing Time Delays in Blood Glucose Regulation for T1DM Using a PD Controller and Smith Predictor Method(Institute of Electrical and Electronics Engineers Inc., 2025-06-27) Kurnaz, Ender; Aydogdu, OmerDiabetes Mellitus, particularly Type 1 Diabetes (T1DM), is a condition where the pancreas does not produce enough insulin, causing abnormal Blood Glucose (BG) levels. Proper BG regulation is essential for managing T1DM and preventing related complications. In this research, a hybrid method that integrates the PD controller and Smith Predictor is employed to tackle the issue of time delays in blood glucose regulation for T1DM patients. This study compares two control strategies: a Proportional-Derivative (PD) controller and the Smith Predictor, both of which address time delays inherent in glucose monitoring and insulin delivery systems. The PD controller parameters are determined using MATLAB Simulink's PID Tuner, and the system's transfer function was derived through System Identification. The time response analysis of both control methods shows that while their rise times are similar, the Smith Predictor offers a significant reduction in delay and peak times. © 2025 Elsevier B.V., All rights reserved.Article Autonomous Fire Fighting Mission Using Unmanned Aerial Vehicle Image Processing(Tokat Gaziosmanpaşa Üniversitesi, 2023) Düzyol, Kevser; Budak, Serkan; Kurnaz, Ender; Durdu, Akif; Samur, İrfan; Aslanbaş, Metehan; Ayık, EmirhanIn this study, an adaptive Unmanned Aerial Vehicle (UAV) design and software for fire extinguishing mission has been realized. With a water release mechanism mounted on the designed UAV, water is taken from the determined area and discharged to the determined area. While designing the UAV, attention was paid to ensure that it is fast, light, maneuverable and has a high load carrying capacity. The body is designed to be high strength and lightweight. For this reason, the UAV Octo Quad X8 (4 rotors, 8 motors & propellers) has a chassis type. It is aimed for the UAV to perform this task autonomously. The Python software language will be used to determine the area where the water will be discharged and the Haversine formula will be used. As a result of the experiments, the hit rate of the water to be discharged into the center of the 3-meter pool was 90%. Bu çalışmada, yangın söndürme görevi için uyarlanabilir bir İnsansız Hava Aracı (İHA) tasarımı ve yazılımı gerçekleştirilmiştir. Tasarlanan İHA’ya monte edilen bir su bırakma mekanizması ile belirlenen bölgeden su alınıp tespit edilen bölgeye suyun boşaltılması sağlanmaktadır. İHA tasarlanırken hızlı, hafif, manevra kabiliyeti yüksek ve yük taşıma kapasitesinin fazla olmasına dikkat edilmiştir. Gövde, yüksek dayanıma sahip ve hafif olacak şekilde tasarlanmıştır. Bu sebepten dolayı İHA Octo Quad X8 (4 rotor, 8 motor&pervane) şase tipine sahiptir. İHA’nın bu görevi otonom bir şekilde yapması amaçlanmıştır. Suyun boşaltılacağı alanın tespiti için Python yazılım dili kullanılacak olup Haversine formülünden yararlanılacaktır. Yapılan denemeler sonucunda 3 metrelik havuzun merkezine boşaltılacak suyun isabet oranı %90 olmuştur.Conference Object Citation - WoS: 1Citation - Scopus: 8Pancreas Segmentation in Abdominal Ct Images With U-Net Model(IEEE, 2020) Kurnaz, Ender; Ceylan, RahimePancreas is one of the most challenging organs in segmentation due to its different shape, position and size in each human being. With the development of machine learning, various deep learning methods are applied to segment the pancreas among organs in the abdominal region. In this study, pancreas segmentation is performed using the U-Net model, which is one of the convolutional neural networks (CNN) models. The results of pancreas segmentation performed on the Pancreas CT data set obtained from The Cancer Imaging Archive (TCIA) database containing computed tomography images of 82 patients are presented in detail. According to the results, Dice similarity coefficient and Jaccard similarity coefficient are found to be 0.78 and 0.66, respectively.Publication Pancreas Segmentation in Abdominal CT Images with U-Net Model(2020) Kurnaz, Ender; Ceylan, RahimePancreas is one of the most challenging organs in segmentation due to its different shape, position and size in each human being. With the development of machine learning, various deep learning methods are applied to segment the pancreas among organs in the abdominal region. In this study, pancreas segmentation is performed using the U-Net model, which is one of the convolutional neural networks (CNN) models. The results of pancreas segmentation performed on the Pancreas CT data set obtained from The Cancer Imaging Archive (TCIA) database containing computed tomography images of 82 patients are presented in detail. According to the results, Dice similarity coefficient and Jaccard similarity coefficient are found to be 0.78 and 0.66, respectively.Master Thesis Abdomen Bt Görüntülerinde Pankreas Segmentasyonu için Yeni Bir Derin Öğrenme Yaklaşımı: Pascal U-net(Konya Teknik Üniversitesi, 2021) Kurnaz, Ender; Ceylan, RahimeGünümüzde derin öğrenme modellerinin medikal görüntü işlemede kullanımı hız kazanmıştır. Özellikle kesit görüntülerinden organ segmentasyonu üzerine gerçekleştirilen çalışmalarda derin öğrenme yöntemleri sıklıkla tercih edilmektedir. Abdomen bölgesinde yer alan pankreas, her insanda şekil, konum ve büyüklük bakımından farklı olduğundan segmentasyonu oldukça zorlayıcıdır. Bu problemin çözümünde literatürde genellikle derin öğrenme modellerinden biri olan U-Net modeli tercih edilmektedir. Bu tez çalışmasında, pankreas segmentasyonu için Pascal üçgenindeki sayı dizilimine uygun bir mimariye sahip ve U-Net modelini temel alan yeni bir derin öğrenme modeli önerilmiştir. Önerilen bu model Pascal U-Net modeli olarak isimlendirilmiştir ve modelin başarımı iki farklı veri seti üzerinde değerlendirilmiştir. İlk olarak halka açık ve literatürde sıklıkla kullanılan bir veri seti olan The Cancer Imaging Archive Pankreas-BT veri setinden yararlanılmıştır. Ayrıca ikinci veri seti olarak Selçuk Üniversitesi Tıp Fakültesi Hastanesi Radyoloji Bölümü'nden alınan abdomen BT görüntüleri kullanılmıştır. Veri setlerindeki kayıtlardan her hasta için bir kesit görüntüsü seçilmiş ve önişleme yöntemleri uygulanarak derin öğrenme ağları için veri setleri oluşturulmuştur. Pascal U-Net modeli ile her iki veri seti üzerinde elde edilen pankreas segmentasyon sonuçlarının karşılaştırılması için, aynı veri setleri üzerinde U-Net modeli ile de segmentasyon işlemi gerçekleştirilmiştir. 2, 4 ve 6 katlı çapraz doğrulama ve 1'den 10'a kadar farklı yığın sayılarında çalıştırılan derin öğrenme modelleri sonucunda elde edilen segmentasyon haritaları, 7 farklı performans metriği kullanılarak değerlendirilmiştir. Her bir yığın sayısı ve farklı kat çapraz doğrulama ile gerçekleştirilen pankreas segmentasyonu sonuçları, 10 kez çalıştırma sonuçlarının ortalamasıdır. Hem U-Net hem de Pascal U-Net segmentasyon sonuçları 7 farklı metrik ve görsel değerlendirmeler temel alınarak analiz edilmiştir. Sonuçlar incelendiğinde; her iki veri setinde de Pascal U-Net modeli, geleneksel U-Net mimarisine karşı Dice Benzerlik Katsayısı metriği bakımından yaklaşık %1'lik bir değer ile üstünlük göstermiştir.Article Closed-Loop Control of Blood Glucose Levels in Simulated Type-1 Diabetes Patients With a New Type of I-PD Controller(Institute of Electrical and Electronics Engineers Inc., 2026) Kurnaz, Ender; Aydoğdu, ÖmerAccurate Blood Glucose (BG) control in Type 1 Diabetes Mellitus (T1DM) remains difficult due to the nonlinear and time-delayed characteristics of the glucose–insulin system. In this study, new type Integral-PD (I-PD) control method, which provides better operating performance in time-delayed systems, will be discussed and its performance in BG control of T1DM patients will be examined. This study evaluates four different model-based control schemes as I-PD, Proportional-Integral-PD (PI-PD), Proportional-Derivative (PD) and Self-Tuning PD (STPD), to compare their performances for closed-loop BG regulation using the UVA/Padova T1DM Simulator. While linearized patient models were derived through System Identification (SI) solely for controller synthesis, the performance and robustness of the proposed algorithms were rigorously validated using the original non-linear dynamics of the UVA/Padova simulator across a 48-hour meal scenario involving 30 virtual subjects. In glucose control applications, linearized patient models were obtained through System Identification (SI) in MATLAB and tested over a 48-hour meal scenario for 30 virtual subjects across three age groups. Results show that I-PD and PI-PD controllers achieved superior regulation with Time in Range (TIR) values of 76.2% and 76.4% in adults, respectively, outperforming PD (65.2%) and STPD (59.4%). In adolescents, TIR remained above 67 %, demonstrating strong robustness and low variability. Pediatric simulations yielded lower stability (TIR <inline-formula> <tex-math notation="LaTeX">$\approx ~51$ </tex-math></inline-formula>–53 %) due to rapid metabolic dynamics. Overall, the findings confirm that integral-assisted PD controllers achieve stable, accurate, and safe BG regulation in closed-loop operation. The proposed control structures demonstrated strong robustness, low variability, and effective mitigation of both hypoglycemia and hyperglycemia, indicating their potential suitability for future integration into artificial pancreas systems.
Research Topics
Domains
Physical SciencesHealth Sciences
Fields
Computer ScienceHealth ProfessionsMedicineEngineering
Subfields
Computer Vision and Pattern RecognitionHealth Information ManagementRadiology, Nuclear Medicine and ImagingAerospace EngineeringEndocrinology, Diabetes and Metabolism
Specific Research Areas
Robotic Path Planning Algorithms
Artificial Intelligence in Healthcare
COVID-19 diagnosis using AI
Advanced Neural Network Applications
Robotics and Sensor-Based Localization
Diabetes Management and Research
Sustainable Development Goals
3GOOD HEALTH AND WELL-BEING
4
Research Products
Documents
5
Citations
30
h-index
2
Documents
4
Citations
16
Publication Collaboration
| Affiliation Name | Count |
|---|---|
| Konya Technical University | 7 |
| Selçuk University | 1 |
| Karamanoğlu Mehmetbey University | 1 |
1 / 1
Data obtained from OpenAlex
| Journal | Count |
|---|---|
| 2018 INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND DATA PROCESSING (IDAP) | 1 |
| 2020 28th Signal Processing and Communications Applications Conference (SIU) | 1 |
| 2020 28TH SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU) | 1 |
| -- 9th International Symposium on Innovative Approaches in Smart Technologies, ISAS 2025 -- Gaziantep -- 211342 | 1 |
| IEEE Access | 1 |
Current Page: 1 / 2
Scholarly Output
9
Articles
4
Views / Downloads
19/11
Supervised MSc Theses
1
Supervised PhD Theses
0
WoS Citation Count
16
Scopus Citation Count
30
Patents
0
Projects
0
WoS Citations per Publication
1.78
Scopus Citations per Publication
3.33
Open Access Source
4
Supervised Theses
1
Scopus Quartile Distribution
Competency Cloud
