Ege University Institutional Repository
Not a member yet
116239 research outputs found
Sort by
İzmir Körfezi'nde bulunan Ulva sp. biyokütlesinden biyogaz eldesi ve biyogaz verimliliğinin belirlenmesi
Full text linkGiderek artan nüfus, sanayileşme ve tekonolojik gelişmeler insanlığın enerjiye olan ihtiyacının giderek artmasına sebep olmuştur. Yenilenemez enerji kaynaklarının yarattığı sorunlar ve tükeniyor oluşu insanlığı yenilenebilir enerji kaynaklarına yöneltmektedir. Bu enerji kaynaklarından bazıları güneş, rüzgar jeotermal ve biyokütle olarak gösterilmektedir. Yenilenebilir enerji kaynaklarından biri olan biyokütle enerjisi, organik maddelerden enerji elde etme potansiyeli ile dikkat çekmektedir. Her ne kadar günümüzde bazı kısıtlamalara sahip olsa bile umut vaadeden bir enerji kaynağıdır. Organik, belediye ve tarım atıkları gibi hammaddeleri kullanarak enerji üretimine katkı sağlamaktadır. Bu hammaddelerden önem seviyesinde yukarılara tırmanan makroalglerdir. Denizel ortamlarda olmaları sebebiyle karasal tarıma rakip değildir. Deniz ekosisteminin vazgeçilmezi olan makroalgler, bulundukları ortamda fazla besine sahip olduklarında hızla çoğalarak alg patlamalarına neden olur. Bu alg patlamalarının fayda ve zararları bulunmakla beraber biyokütle enerjisi için önemli bir hammadde kaynağıdır. Bu tezin amacı İzmir Körfez'inde alg patlamasıyla çevresel sorunlara yol açan ve atık olan Ulva lactuca makroalginin biyogaz üretim potansiyelini incelemek ve dönemsel farklılıkların incelemektir. Biyokütlelerin sürekli beslemeli gaz üretim denemeleri sürekli karıştırmalı tank biyoreaktörde gerçekleştirilmiştir. Çalışma sonucunda toplanan biyokütlenin dönemsel farkları karşılaştırılmış ve bir ön yıkama işleminin farkı olmadığını ortaya koymuştur.The increasing population, industrialization, and technological advancements have led to a growing demand for energy. The problems caused by non-renewable energy sources and their depletion have directed humanity towards renewable energy sources. Some of these energy sources include solar, wind, geothermal, and biomass energy. Among these, biomass energy stands out for its potential to generate energy from organic materials. Although it currently faces some limitations, it is a promising energy source. By utilizing raw materials such as organic waste, municipal waste, and agricultural residues, it contributes to energy production. One raw material that has been gaining importance is macroalgae. As they inhabit marine environments, they do not compete with terrestrial agriculture. Macroalgae, which are indispensable to the marine ecosystem, can rapidly proliferate when there is an excess of nutrients in their environment, causing algal blooms. While these blooms have both benefits and drawbacks, they are a significant raw material source for biomass energy. The aim of this thesis is to examine the biogas production potential of Ulva lactuca, a macroalga causing environmental issues in the İzmir Bay through algal blooms, and to investigate seasonal variations. Continuous feeding gas production trials of the biomass were conducted in a continuously stirred tank bioreactor. The study compared the seasonal differences of the collected biomass and revealed that a pre-washing process did not make a significant difference
The Problem of Algorithmic Minds, Theological Critique of Mind Transfer
No full textIn the process of technological integration, digitalization, genetics, cybernetics, neuroscience and finally general artificial intelligence, our practices and our system of values, which are fed by religion and tradition, are undergoing a rapid metamorphosis. By its foundational nature, the science of kalam produces thought not only on problematic matters but also on theoretical subjects. In this context, generating ideas about mind transfer and its possible consequences, which is one of the topics currently under active discussion, is among the intellectual responsibilities of kalam. Mind transfer is unlikely in today's conditions, but the dizzying progress in AI technologies has the potential to open up different opportunities in the future. For example, it is suggested that in the coming decades, the human self could be neuronally mapped and turned into a mathematical algorithm or subject to biochemical transfer. Subsequently, electronic personalities and neurochemical maps could be transferred to robots called humanoids through neuroengineering or chip technologies. This article will first examine the scientific and philosophical evidence for the possibility of an algorithmic mind and then discuss the theological problems that may arise if this project is realized. It is concluded that even if mind transference remains a mere thought experiment, the intellectual endeavor in this topic will offer new neurotheological perspectives on the relationship between the self and the brain. The article will also discuss the theological implications of the realization of mind transfer. In this study, which we approached with the idea that it would force us into a new theology, literature review, descriptive analysis, content analysis and qualitative research methods were utilized. It is hoped that the article, which we notice that no research has been conducted from a theological perspective before, will contribute to the field as it is the first of its kind
Analyzing the Impact of Transmission Strategies on Localization Performance in Wireless Sensor Networks
No full textLocalization, essential in WSN applications, enables sensor nodes to determine their physical positions by referencing anchor nodes. We evaluate broadcast and unicast packet transmissions at the data-link layer for their impact on localization performance. Implemented on the Contiki-NG operating system, the study examines how anchor node density and antenna range affect localization success and the number of required anchor nodes between broadcast-based and unicast-based localization propagation in protocol stack. Results using Cooja simulator, demonstrate the trade-offs between unicast and broadcast transmission approaches, particularly in terms of network overhead, energy consumption and localization performance. For instance, with an antenna range of 20 meters, achieving a localization ratio of over 90% requires only 20% anchor density with broadcast transmission, whereas unicast transmission requires a 60% anchor density to achieve the same ratio. This demonstrates that broadcast localization can lead to approximately a 33% reduction in hardware costs, offering significant efficiency gains. These findings provide insights into optimal propagation techniques and highlight the advantages of broadcasting in resource-constrained WSN deployments
Evaluation of nisin-added adhesive resins: antimicrobial properties, dentin bond strength, degree of monomer conversion and structural/chemical characterization
No full textThis study evaluated the antibacterial effect, dentin bond strength, monomer conversion degree, and chemical structure of an adhesive resin containing nisin, an antimicrobial peptide (AMP) derived from Lactococcus lactis. Antimicrobial activity against S. mutans was tested using a disk diffusion method with various nisin concentrations (1:1 to 1:80). The optimum concentration, 1:40, was selected for further testing. The adhesive was characterized using XRD, EDX, FT-IR, and SEM analyses. For bond strength testing, 80 caries-free molars with enamel removed were divided into control and AMP groups, each further subdivided into etch-and-rinse or self-etch application subgroups (n = 20). FT-IR was used to measure monomer conversion degrees, and data were statistically analyzed. Antimicrobial testing showed inhibition zones of 27 mm for nisin alone, 9 mm for resin, and 12 mm for resin with 1:40 nisin. SEM and EDX analyses confirmed homogeneous nisin distribution, while XRD indicated no structural changes to the adhesive. In the etch-and-rinse mode, the AMP group exhibited higher bond strength compared to the Control group though the difference was not statistically significant (p > 0.5). Similarly, no significant differences were observed between the groups in the self-etch mode (p > 0.5). When comparing the bond strengths of adhesive resins according to the application protocol, no significant differences were found between the etch-and-rinse and self-etch modes for either adhesive (p > 0.5). The experimental adhesive exhibited a lower monomer conversion degree than the control group (p < 0.05). The nisin-containing experimental adhesive showed antimicrobial activity against S. mutans, preserved bond strength in etch-and-rinse and self-etch mode, and maintained its structural integrity.NEVU Scientific Research Projects Coordination; [GAP21SAG3]This study was supported by NEVU Scientific Research Projects Coordination (GAP21SAG3). No potential conflict of interest was reported by the author(s)
EGEFACE: A new face memory test with static and dynamic images
No full textFace memory is a crucial cognitive ability necessary for maintaining a healthy social life. Recent studies reveal large individual differences in face recognition ability. Face memory tests are used to evaluate this ability. The main purpose of this study was to develop a new face memory test (EGEFACE) addressing the limitations of existing tests using both static and dynamic stimuli to increase ecological validity; employing face recognition algorithms to adjust test difficulty; measuring face memory accuracy independently of response bias by including both target-absent and target-present trials and using ROC analysis; and developing a test to measure both ends of the face recognition ability spectrum. After building a new database of static and dynamic faces, we created three difficulty levels using a face recognition algorithm. We collected data from 703 participants in two steps and examined the internal consistency, split-half reliability, and item–total score correlations. The reliability analysis confirmed that both target-absent and target-present trials of EGEFACE were reliable. High EGEFACE performers scored near super recognizer levels on CFMT+, while low performers showed limited overlap with prosopagnosic-level performance on CFMT+, suggesting EGEFACE’s sensitivity across different levels of face recognition ability. Overall, results indicated a moderate positive correlation between EGEFACE and CFMT+, showing that both tests assess similar cognitive skills, while a low to moderate correlation with KFMT suggests that EGEFACE measures cognitive ability that is related to yet distinct from face perception. The results suggest that EGEFACE shows promise as an ecologically valid and effective alternative tool for assessing individual differences in face memory. © 2025 Elsevier B.V., All rights reserved
Safety and efficacy of a fitusiran antithrombin-based dose regimen in people with hemophilia A or B: the ATLAS-OLE study
No full textFitusiran, a subcutaneous investigational small interfering RNA therapeutic, lowers antithrombin (AT) to increase thrombin generation and rebalance hemostasis in people with hemophilia. This phase 3 open-label extension study (ATLAS-OLE) evaluated safety and efficacy of an AT-based dose regimen (AT-DR) in males aged ?12 years with severe hemophilia A/B, with/without inhibitors. The original dose regimen (ODR) of 80 mg monthly was optimized to AT-DR targeting AT activity levels 15% to 35% to mitigate thrombotic risk (starting dose of 50 mg once every 2 months, individually adjusted to 20 mg once every 2 months, or 20/50/80 mg monthly as needed). Primary and secondary end points were safety and efficacy, respectively. Integrated safety analyses assessed safety of AT-DR and ODR across all fitusiran studies and integrated efficacy analyses compared efficacy of AT-DR in ATLAS-OLE with phase 3 parent study control groups. At interim data cutoff, 213 participants were enrolled on AT-DR (78% on regimens of once every 2 months). Integrated safety analyses of participants receiving AT-DR (n = 286) demonstrated that AT-DR was well tolerated. In ATLAS-OLE, median observed annualized bleeding rate (ABR) with AT-DR was 3.7 (interquartile range, 0.0-7.5). Integrated efficacy analyses demonstrated superiority of AT-DR over on-demand clotting factor concentrates (CFCs; 71% mean ABR reduction; P < .0001), and on-demand bypassing agents (BPAs; 73% mean ABR reduction; P = .0006); improvement over BPA prophylaxis (70% mean ABR reduction); and ABR comparable with that observed with CFC prophylaxis. Fitusiran AT-DR was well tolerated and maintained bleed protection with as few as 6 injections per year. This trial was registered at www.ClinicalTrials.gov as #NCT03754790. © 2025 Elsevier B.V., All rights reserved
Radyonüklid işaretli manyetik nanoparçacıkların ve türevlerinin sentezlenmesi ve teranostik potansiyellerinin araştırılması
No full textBu tez çalışmasında, hedefe yönelik tedavi için radyonüklid işaretli manyetik nanoparçacık sistemlerinin geliştirilmesi amaçlanmıştır. HER2 reseptör pozitif kanser hücrelerini hedeflemek üzere, trastuzumab antikoru ve [19F]FDG-(2-deoksi-2-[florin-19]floro-D-glukoz) (FDG) molekülü ile konjuge edilmiş, yüzey modifikasyonları optimize edilmiş kübik demir oksit nanoparçacıkları tasarlanmıştır. Çalışma kapsamında, nanoparçacık sentezinden sistemin biyolojik etkinliğinin değerlendirilmesine kadar tüm süreçler sistematik bir şekilde gerçekleştirilmiştir. İlk aşamada, kübik demir oksit nanoparçacıkları (C-Fe3O4) başarıyla sentezlenmiştir. Bu nanoparçacıkların yüzey stabilitesini ve fonksiyonelliğini artırmak amacıyla, nanoparçacık yüzeyleri sırasıyla tetraetil ortosilikat (TEOS), (3-aminopropil)trietoksisilan (APTES) ve polietilen glikol (PEG) kullanılarak modifiye edilmiştir (C-Fe3O4-SiO2-NH2-PEG). Yüzey modifikasyonu tamamlanan nanoparçacıklar, FDG molekülü ile konjuge edilerek tümör spesifikliği ve biyouyumluluğu optimize edilmiştir (C-Fe3O4-SiO2-NH2-FDG ve C-Fe3O4-SiO2-NH2-PEG-FDG). Reseptör spesifitesi, hem C-Fe3O4-SiO2-NH2-PEG-FDG hem de C-Fe3O4-SiO2-NH2-FDG nanoparçacıkları için ayrı ayrı değerlendirilmiştir. Yapılan in vitro testler sonucunda, C-Fe3O4-SiO2-NH2-FDG nanoparçacıklarının daha yüksek tutulum gösterdiği belirlenmiş ve bu nedenle hücre deneylerinde bu nanoparçacıkların kullanılması tercih edilmiştir. Ardından, trastuzumab antikoru, p-NCS-Bz-DOTA (DOTA) şelatörü ile konjuge edilmiş ve bu yapı, optimize edilmiş koşullarda lutesyum-177 (177Lu) ile radyoişaretlenmiştir (177Lu-DOTA-Tras). Son olarak 177Lu-DOTA-Tras, modifiye edilmiş FDG konjuge manyetik nanoparçacıklarla birleştirilerek hedefe yönelik terapötik bir sistem oluşturulmuştur (177Lu-DOTA-Tras@C-Fe3O4-SiO2-NH2-FDG). Geliştirilen nanoparçacık sisteminin biyolojik etkinliği, HER2 pozitif SKOV-3 (yumurtalık) ve BT474 (triple pozitif meme kanseri) hücre hatlarında in vitro olarak test edilmiştir. Radyokonjugat, HER2 pozitif kanser hücrelerinde 20 MBq/mL dozda 48 saatlik inkübasyon sonrası radyoaktif olmayan kontrollere kıyasla belirgin bir şekilde hücre canlılığını azaltmıştır. Radyokonjugatın hedef hücrelerde yüksek spesifik alımı ve belirgin radyoterapi etkinliği gözlemlenmiştir. Triple negatif MDA MB231 meme kanseri hücreleri, sistemin seçiciliğini doğrulamak amacıyla negatif kontrol grubu olarak kullanılmış ve bağlanma deneylerinde herhangi bir spesifik etki gözlenmemiştir. Bu sonuçlar, sistemin hedefe yönelik ve seçici bir terapötik ajan olarak etkinliğini ortaya koymaktadır. Sonuç olarak, bu tez çalışması, HER2 pozitif kanserlerin tedavi ve tanısında kullanılabilecek, biyouyumlu ve düşük yan etkili radyonüklid işaretli manyetik nanoparçacıkların geliştirilmesiyle, kanser tedavisinde yeni nesil teranostik yaklaşımlara önemli bir katkı sunmaktadır.In this thesis, the development of radiolabeled magnetic nanoparticle systems for targeted therapy has been aimed. To target HER2 receptor-positive cancer cells, surface-modified cubic iron oxide nanoparticles conjugated with trastuzumab antibody and [19F]FDG-(2-deoxy-2-[fluorine-19]fluoro-D-glucose) (FDG) molecule were designed. Throughout the study, all processes, from nanoparticle synthesis to the evaluation of the system9s biological activity, were systematically carried out. In the initial stage, cubic iron oxide nanoparticles (C-Fe3O4) were successfully synthesized. To enhance the surface stability and functionality of these nanoparticles, their surfaces were modified sequentially using tetraethyl orthosilicate (TEOS), (3-aminopropyl)triethoxysilane (APTES), and polyethylene glycol (PEG), resulting in C-Fe3O4-SiO2-NH2-PEG. The surface-modified nanoparticles were then conjugated with the FDG molecule to optimize tumor specificity and biocompatibility (C-Fe3O4-SiO2-NH2-FDG and C-Fe3O4-SiO2-NH2-PEG-FDG). Receptor specificity was evaluated separately for both C-Fe3O4-SiO2-NH2-PEG-FDG and C-Fe3O4-SiO2-NH2-FDG nanoparticles. In vitro tests revealed that C-Fe3O4-SiO2-NH2-FDG nanoparticles exhibited higher uptake; therefore, these nanoparticles were preferred for further cell studies. Subsequently, the trastuzumab antibody was conjugated with the p-NCSBz-DOTA (DOTA) chelator, and this structure was radiolabeled with lutetium-177 (177Lu) under optimized conditions, forming 177Lu-DOTA-Tras. Finally, 177Lu-DOTA-Tras was combined with the modified FDG-conjugated magnetic nanoparticles to develop a targeted therapeutic system (177Lu-DOTA-Tras@CFe3O4-SiO2-NH2-FDG). The biological efficacy of the developed nanoparticle system was evaluated in vitro using HER2-positive SKOV-3 (ovarian) and BT474 (triple-positive breast cancer) cell lines. The radioconjugate significantly reduced cell viability in HER2-positive cancer cells after a 48-hour incubation at a dose of 20 MBq/mL, compared to non-radioactive controls. The radioconjugate demonstrated high specific uptake in target cells and notable radiotherapy efficacy. Triple-negative MDA MB231 breast cancer cells were used as a negative control group to verify system selectivity, and no specific effect was observed in binding experiments. These results highlight the system9s effectiveness as a targeted and selective therapeutic agent. In conclusion, this thesis contributes significantly to next-generation theranostic approaches in cancer treatment by developing biocompatible and lowtoxicity radiolabeled magnetic nanoparticles for the diagnosis and treatment of HER2-positive cancers
Polyacrylonitrile Nanofibers Containing Microencapsulated Phase Change Materials Produced for Thermal Energy Storage by Electrospinning Fabric
No full textYMO150219B09In this study, microencapsulated phase change materials (MEPCM) were imparted into nanofibers using electrospinning method and an original material composite systems, microparticles interconnected through nanofibers were produced. Poly(methyl methacrylate-co-ethylene glycol dimethacryate-co-2-hydroxyethyl acrylate) and poly(methyl methacrylate-co-ethylene glycol dimethacrylate-co-2-hydroxyethyl methacrylate) shells and n-octadecane and Rubitherm RT-21 core were formed for this study. Composites with low diameter nanofibers as compored to microparticles consisted of polyacrylonitrile (PAN) nanofibers and dispersed MEPCMs together. MEPCMs were produced through commonly used emulsion polymerization technique. They were dispersed into PAN solution in dimethylformamide (DMF) and electrospinned. Thermophysical properties of MEPCMs dispersed nanofibers were determined using differential scanning calorimetry (DSC) as Fourier Transform Infrared (FTIR) Spectroscopy instrument was used to prove copresence of core, shell, and nanofiber materials. Besides, Scanning Electron Microscopy (SEM) device was used to examine the morphology of MEPCMs and the MEPCM nanofiber composites.Çankırı Karatekin Üniversitesi BAP Birim Koordinatörlüğ
Üniversite öğrencilerinin iklim değişikliğine yönelik algı ve kaygı düzeyleri ile davranış biçimlerinin analizi
No full textSanayi Devriminden itibaren artan fosil yakıt kullanımı, ormansızlaştırma, kentleşme ve arazi kullanım değişikliklerine bağlı olarak meydana gelen antropojenik iklim değişikliği; ekolojik, çevresel, sosyo-kültürel ve ekonomik sistemleri olumsuz etkilemekte ve sağlık sorunları, gıda güvenliği, su kıtlığı, biyoçeşitlilik kayıpları, orman yangınları ve iklim mülteciliği gibi iklim değişikliğiyle ilişkili birçok risk unsurunu gündeme getirmektedir. Günümüzün en önemli sorunlarının başında gelen ve rasyonel adımların atılmaması durumunda yaşamın birçok alanını doğrudan tehdit eden antropojenik iklim değişikliğiyle mücadelenin başarıya ulaşması için, sürecin ilk aşaması olan bireysel algı, kaygı, tutum ve davranışların anlaşılması son derece önem arz etmektedir. Tez çalışmasının amacı, üniversite öğrencilerinin iklim değişikliğine ilişkin algı ve kaygı düzeyleri ile davranış biçimlerini ortaya koymaktır. Bu kapsamda araştırmaya katılan bireylerin sosyo-demografik profillerine göre iklim değişikliğine yönelik algı, kaygı ve davranışları arasındaki gruplararası farklılaşmanın boyutlarının ortaya konulması ve bu farklılaşmanın en fazla görüldüğü değişkenlerin belirlenmesi çalışmanın ana hedefleri arasındadır. Araştırmanın örneklemi, Ege Üniversitesinin çeşitli bölümlerinde öğrenim gören öğrencilerden oluşmaktadır. Bu kapsamda tez çalışması, nicel araştırma yöntemleri kapsamında yürütülmüş olup 409 katılımcıya yüz yüze anket uygulanmış ve katılımcıların iklim değişikliğine ilişkin algı, kaygı ve davranışları çeşitli analizler vasıtasıyla incelenmiştir. Elde edilen bulgular; bireylerin sosyo-demografik özelliklerine göre iklim değişikliğine yönelik algı, endişe ve davranışlarında önemli farklar görüldüğü; özellikle dünya görüşü, doğayla kurulan ilişkiler, iklim değişikliğine yönelik doğru ve yeterli bilgi düzeyi ve iklim değişikliği ilgisinin bireylerin algı, kaygı ve davranış biçimlerinin şekillenmesinde oldukça etkili olduğuna ve gruplararası anlamlı farklılaşmaların bulunduğuna işaret etmiştir. Bireylerin iklim değişikliğine yönelik algı, tutum ve davranışlarını en fazla etkileyen değişkenlerin; siyasi görüş, dindarlık düzeyi, kırsal arka plan ve aktif gönüllülük eğilimleri olduğu saptanmıştır. Özellikle sol ideolojik görüşte olan ve doğayla güçlü ilişkiler kuran seküler öğrenciler, sağ görüşteki muhafazakâr kentli katılımcılara kıyasla iklim değişikliğini daha yüksek bir risk olarak algıladıkları ve daha fazla endişeli oldukları görülürken mücadeleye ilişkin davranış sergileme ve gelecekte davranışlarını iklim değişikliğiyle mücadeleye uygun olacak biçimde değiştirme eğilimi göstermektedir.Since the Industrial Revolution, anthropogenic climate change, driven by fossil fuel consumption, deforestation, urbanization, and land-use changes, has adversely affected ecological, environmental, socio-cultural, and economic systems. This phenomenon has brought numerous climate-related risk factors to the forefront, including health problems, food security challenges, water scarcity, biodiversity loss, wildfires, and also known as climate refugee climate-induced migration. As one of the most pressing issues of our time, anthropogenic climate change poses a significant threat to various aspects of life if rational measures are not taken. For efforts to combat climate change to be successful, it is crucial to understand individual perceptions, attitudes, and behaviors, which constitute the initial phase of this process. The primary aim of this dissertation is to examine university students' perceptions, levels of concern, and behavioral patterns. In this context, the study seeks to identify the extent of intergroup differences in climate change perception, anxiety, and behavior based on the demographic profiles of the participants. Additionally, determining the variables in which these differences are most pronounced constitutes a key objective of the research. The sample of this study consists of university students from various departments at Ege University. Within this scope, the research was designed using quantitative methods, and a survey was conducted with 409 participants. The findings indicate significant differences in individuals' perceptions, attitudes, and behaviors toward climate change based on their socio-demographic characteristics. Notably, worldview, relationships with nature, the accuracy and adequacy of knowledge about climate change, and the level of interest in the issue play a crucial role in shaping individuals' perceptions, concerns, and behavioral patterns, leading to statistically significant intergroup differences. Among the key variables influencing individuals' perceptions, attitudes, and behaviors toward climate change, political worldview, level of religiosity, rural background, and active volunteerism tendencies stand out. One of the study's significant findings is that secular students with leftleaning ideological views and strong connections to nature perceive climate change as a greater risk and exhibit higher levels of concern compared to conservative, rightleaning urban participants. Moreover, they demonstrate a greater willingness to engage in climate action and adapt their future behaviors in alignment with climate change mitigation efforts
YOLOv8 ve SAM derin öğrenme modellerinin beyin tümörü tespiti ve segmentasyonu için MR görüntülerinde değerlendirilmesi ve entegrasyonu
No full textModern tıp pratiğinde, manyetik rezonans (MR) gibi medikal görüntüleme teknikleri, hastalıkların teşhisi için kritik bir rol oynamaktadır. Bu süreçlerdeki temel adımlardan biri olan görüntü segmentasyonu, anatomik yapıların hassas bir şekilde ayırt edilmesini sağlar. Ancak bu işlemin radyologlar tarafından manuel olarak yapılması, zaman alıcı, yoğun emek gerektiren ve farklı uzmanlar arasında tutarlılık sorunları barındıran bir süreçtir. Yapay zekanın bir dalı olan derin öğrenme, medikal görüntü analizi alanında bu zorluklara güçlü bir çözüm sunmaktadır. Derin öğrenme tabanlı otomatik segmentasyon modelleri, insan gözünün fark edemeyeceği karmaşık örüntüleri öğrenebilir ve segmentasyon işlemini saniyeler içinde, yüksek doğruluk ve tekrarlanabilirlik ile gerçekleştirebilir. Bu nedenlerle, bu tez çalışmasında derin öğrenme yaklaşımıyla beyin tümörü (menenjiyom, gliom ve hipofiz) MR görüntülerinin otomatik segmentasyonu için güncel modellerin eğitilmesi, test edilmesi ve karşılaştırılması amaçlanmıştır. Çalışmada, 233 hastaya ait 3064 adet T1 ağırlıklı kontrastlı MR görüntüsü içeren, halka açık bir veri seti kullanılmıştır. YOLOv8-Segmentasyon modeli, bu veri seti üzerinde eğitilmiştir. Model hattı için ise önce bir YOLOv8 tespit modeli, tümörlerin etrafında sınırlayıcı kutular üretmek üzere eğitilmiştir. Bu kutular daha sonra YOLOv8-SAM ve YOLOv8-MedSAM model hatlarında sırasıyla birer girdi istemi olarak kullanılmıştır. Tüm modellerin nihai performansı, eğitimde hiç karşılaşmadıkları test veri seti üzerinde ve eşit donanım koşulları altında ölçülmüştür. Elde edilen bulgulara göre, göreve özel olarak eğitilmiş YOLOv8- Segmentasyon modeli, hem en yüksek segmentasyon doğruluğunu (genel DSC: 0.789) hem de en hızlı çıkarım süresini (17.72 ms) sunarak en üstün ve dengeli performansı göstermiştir. Model hatları arasında, daha büyük bir mimariye sahip genel amaçlı SAM kullanan YOLOv8-SAM hattı (DSC: 0.776), medikal alana özelleşmiş ancak daha küçük mimarili MedSAM'den (DSC: 0.528) belirgin şekilde daha iyi performans sergilemiştir. Bu sonuç, temel model tabanlı yaklaşımlarda modelin ham kapasitesinin, alan özelleşmesinden daha baskın bir faktör olabileceğini göstermiştir. Ayrıca, tüm modellerin, sınırları en belirsiz olan "gliom" sınıfında en düşük performansı sergilediği gözlemlenmiştir. Farklı derin öğrenme stratejilerinin karşılaştırıldığı bu çalışmanın, ileride benzer konular üzerinde yapılacak araştırmalara yol göstermesi hedeflenmektedir. Elde edilen bulgular, iyi tanımlanmış medikal segmentasyon görevleri için tam denetimli ve göreve özel eğitilmiş modellerin, şu anki teknolojiyle hala en güvenilir ve pratik çözüm olduğunu göstermiştir. Bununla birlikte, tespit ile istemli segmentasyon gibi modüler yaklaşımların sunduğu esneklik, bu alanda gelecekteki çalışmalar için önemli bir potansiyel taşımaktadır.In modern medical practice, imaging techniques such as magnetic resonance imaging (MRI) play a critical role in disease diagnosis. One of the fundamental steps in these processes is image segmentation, which allows for the precise delineation of anatomical structures. However, performing this task manually by radiologists is time-consuming, labor-intensive, and prone to inconsistencies between different experts. Deep learning, a branch of artificial intelligence, offers a powerful solution to these challenges in the field of medical image analysis. Deep learning-based automatic segmentation models can learn complex patterns that are imperceptible to the human eye and perform segmentation within seconds, with high accuracy and repeatability. For these reasons, this thesis aims to train, test, and compare state-of-the-art models for the automatic segmentation of brain tumors (meningioma, glioma, and pituitary) in MRI images using a deep learning approach. In this study, a publicly available dataset containing 3,064 contrast-enhanced T1-weighted MR images from 233 patients was utilized. The YOLOv8- Segmentation model was trained on this dataset. For the model pipeline, a YOLOv8 detection model was first trained to generate bounding boxes around tumors. These boxes were then used as prompt inputs in the YOLOv8-SAM and YOLOv8- MedSAM pipelines, respectively. The final performance of all models was evaluated on a test dataset that had not been seen during training, under identical hardware conditions. According to the findings, the task-specific YOLOv8-Segmentation model demonstrated the best and most balanced performance by achieving both the highest segmentation accuracy (overall DSC: 0.789) and the fastest inference time (17.72 ms). Among the model pipelines, the YOLOv8-SAM pipeline, which used the general-purpose SAM with a larger architecture, performed significantly better (DSC: 0.776) than the MedSAM, which, although specialized for the medical domain, had a smaller architecture (DSC: 0.528). This result suggests that in foundation model-based approaches, the raw capacity of the model may be a more dominant factor than domain specialization. Additionally, it was observed that all models exhibited the lowest performance in the <glioma= class, which has the most ambiguous boundaries. This study, in which different deep learning strategies are compared, aims to guide future research on similar topics. The findings indicate that for well-defined medical segmentation tasks, fully supervised and task-specific trained models still represent the most reliable and practical solution with current technology. Nevertheless, the flexibility offered by modular approaches such as detection followed by prompt-based segmentation holds significant potential for future studies in this field