Основные понятия
Proposing novel dual structure-aware image filterings (DSAIF) for semi-supervised medical image segmentation to improve segmentation performance by leveraging topological structure.
Аннотация
The content discusses the importance of leveraging unlabeled images in training for semi-supervised medical image segmentation. It introduces the concept of dual structure-aware image filterings (DSAIF) to preserve topological structure while enhancing appearance diversity. The proposed method significantly outperforms state-of-the-art methods on benchmark datasets.
Статистика
Extensive experimental results on three benchmark datasets demonstrate the proposed method significantly outperforms state-of-the-art methods.
Achieves 90.63% and 91.63% Dice coefficient using 10% and 20% labeled data, respectively.
Using 20% labeled data achieves ∼99.8% Dice performance of using full dataset.
Цитаты
"Most methods maintain consistent predictions of the unlabeled images under variations in the image and/or model level."
"The proposed DSAIF significantly boosts the performance of baseline models."