Fig. 6

Bar graph with a scatter plot overlay for different methods on test samples. (a) and (b) respectively represent the results trained by 10% labeled data and 20% labeled data on the LA dataset. From left to right, the methods compared are UAMT [38], SASSNet [55], LG-ER-MT [56], DTC [36], and the proposed method. In (a), only the mean value of the UA-MT method with 10% labeled data is shown in the bar graph since the model files were not publicly available. (c) and (d) display the results trained with 10% and 20% labeled data on the left ventricle LV dataset. From left to right, the methods are UAMT, DTC, and the proposed method