Annotation variability remains a substantial challenge in medical image segmentation, stemming from ambiguous imaging boundaries and diverse clinical expertise. Traditional deep learning methods producing single deterministic segmentation predictions often fail to capture these annotator biases. In this study, we propose DiffOSeg, a two-stage diffusion-based framework, which aims to simultaneously achieve both consensus-driven (combining all experts' opinions) and preference-driven (reflecting experts' individual assessments) segmentation.
When I first entered the field of medical image analysis, I learned that MICCAI is one of the most prestigious conferences in this area. During my research and studies, papers published at MICCAI have been a great source of inspiration for me. Having my own paper accepted at MICCAI has always been a dream of mine. I am truly honored that this dream has finally come true in Daejeon, 2025!
I will be exploring many topics during the conference. My main interests include uncertainty quantification in segmentation, multi-modal fusion methods, medical vision-language foundation models and so on.
While I'm at the conference, I'm looking forward to attending workshops to learn from leading experts. I hope to actively engage with researchers during poster sessions and networking events. I'm also excited to experience the international academic atmosphere and gain new inspiration from diverse perspectives. Lastly, I'd love to explore the local culture and attractions in Daejeon.