Akademik Veri Yönetim Sistemi
33. IEEE Sinyal İşleme ve İletişim Uygulamaları (SIU), İstanbul, Türkiye, 25 Haziran 2025, (Tam Metin Bildiri)
Accurate bladder volume estimation in the pelvic region diseases play a crucial role in the diagnosis and treatment of diseases. In this study, a You Only Look Once (YOLO)-based model was developed to determine bladder volume. The YOLOv11s-seg segmentation model was employed to delineate bladder boundaries in computed tomography (CT) images. The model was trained on a private dataset comprising five patients, with data augmentation techniques applied to enhance accuracy. The YOLOv11s-seg model demonstrated high-precision segmentation, achieving an accuracy of 97.2%, recall of 83%, and mAP50 of 90.9%. Additionally, three-dimensional (3D) bladder models were generated using segmentation results, and volume calculations were performed. The model's rapid and efficient segmentation capability suggests it as a reliable alternative for bladder volume estimation. Future studies aim to validate the model in clinical settings and improve its accuracy using larger datasets.