核心概念
자가 및 혼합 감독을 통해 다중 클래스 이미지 분할의 훈련 레이블을 효과적으로 개선하는 방법
統計資料
제안된 방법의 실험 결과:
"Validation results on 11 patients showed that the accuracy of training labels was statistically significantly improved, with the Dice similarity coefficient of muscle, subcutaneous and visceral adipose tissue increased from 74.2% to 91.5%, 91.2% to 95.6%, and 77.6% to 88.5%, respectively (p<0.05)."
"The proposed method performed better than the baseline method (Table 1). DSC was significantly improved by 17% for muscle (𝑝 < 0.05), 4% for subcutaneous adipose tissue (𝑝 < 0.05), and 11% for visceral adipose tissue (𝑝 < 0.05)."
引述
"The proposed method was applied to a three-class segmentation of muscle, subcutaneous and visceral adipose tissue on abdominal CT scans."
"Experimental results on multi-class adipose tissue and muscle segmentation showed that the updated labels from proposed method was accurate enough for downstream segmentation models."