Date: 17/06/2026
Hour: 09.00
Student Name & ID: Berke ERSÖZ 255105151
Supervisor: Prof. Dr. Ömer KARAL
Topic: Investigation of CNN Architectures from the Perspective of Biomedical Applications
Link or Room: A516
Abstract:
Convolutional neural networks (CNNs) hold a significant place in the field of deep learning, particularly due to their success in automated feature extraction, high-accuracy classification, image processing, and the analysis of complex data structures. While most features are manually determined by experts in traditional methods, CNN-based models can directly learn meaningful features from data, thus yielding effective results on different data types. This study will evaluate the role of 1D CNNs in the analysis of time series data, ECG, EEG, and other biomedical signals; the use of 2D CNNs in medical image visualization, classification, and segmentation applications; and the importance of 3D CNNs in processing volumetric medical data such as CT and MRI scans. Furthermore, the study will examine how different CNN types perform depending on the input data size, which problems they excel in, and how they can contribute to decision support systems in biomedical fields. Finally, the fundamental building blocks and model development approaches used in CNN models will be explained in general terms.