I am a PhD candidate in Computer Engineering at Gazi University, exploring the frontiers of medical image segmentation and self-distillation. My work focuses on improving model performance and uncertainty estimation in clinical applications. I am passionate about pushing deep learning for healthcare further, developing methods that make AI-assisted diagnosis more reliable and accurate. My current research interests include knowledge distillation and advanced segmentation techniques for medical imaging.
We will present our work on PRISM: Past-Regularized Iterative Self-Distillation with Momentum for Polyp Segmentation at the AE-CAI Workshop, part of the MICCAI 2025 Workshops in Daejeon, South Korea.
We present our work PlutoNet: An Efficient Polyp Segmentation Network with Modified Partial Decoder and Decoder Consistency Training at the AE-CAI Workshop, part of the MICCAI 2024 Workshops in Marrakesh, Morocco.
We present our work An Efficient Polyp Segmentation Network at the 2022 Signal Processing and Communications Applications Conference.
The paper introduces a self-distillation method that uses model uncertainty to enhance polyp segmentation, guiding the network to produce more reliable and consistent predictions.
PlutoNet is a lightweight polyp segmentation model that uses a novel decoder consistency training with a shared encoder, modified partial decoder, and auxiliary decoder to efficiently capture multi-scale and semantic features, achieving superior performance with far fewer parameters than state-of-the-art models.