“…Although recent deep learning based automatic segmentation engines [16,21,29,34] have achieved impressive performance, they still struggle to achieve sufficiently accurate and robust results for clinical practice, especially in the presence of poor image quality (e.g., noise, low contrast) or highly variable shapes (e.g., anatomical structures). Consequently, interactive segmentation [2,18,27,28,32,33] garners research interests of the medical image analysis community, and recently became the choice in many real-life medical applications.…”