Table 1 Comparison with state-of-the-art methods on precision, recall, dice, and Hausdorff distance (HD).

Methods	Pediatric PSAX				Pediatric A4C				EchoNet-dynamic
Methods	Precision	Recall	Dice	HD[1]	Precision	Recall	Dice	HD	Precision	Recall	Dice	HD
UNet (FCN)¹⁸	0.849	0.876	0.862	4.30	0.828	0.834	0.831	4.35	0.884	0.877	0.881	4.41
UNet (PSPNet)¹⁸	0.865	0.870	0.868	4.25	0.820	0.861	0.840	4.30	0.881	0.890	0.885	4.35
PVT¹⁵	0.873	0.837	0.855	4.57	0.867	0.837	0.852	4.61	0.917	0.904	0.910	4.57
UniFormer²⁰	0.910	0.913	0.907	3.52	0.897	0.892	0.892	3.55	0.896	0.916	0.906	3.50
SpectFormer²¹	0.913	0.906	0.916	3.43	0.908	0.904	0.901	3.45	0.920	0.923	0.921	3.40
Flatten Transformer¹⁶	0.922	0.912	0.917	3.35	0.913	0.883	0.898	3.40	0.922	0.928	0.925	3.35
MaxVit¹⁷	0.902	0.925	0.914	3.45	0.891	0.907	0.899	3.50	0.925	0.928	0.926	3.44
U-Mamba⁹	0.928	0.902	0.914	3.45	0.902	0.886	0.894	3.50	0.933	0.916	0.924	3.45
TransUnet¹⁹	0.856	0.806	0.830	5.02	0.827	0.839	0.833	5.05	0.926	0.899	0.908	5.02
Segformer²²	0.865	0.835	0.849	4.80	0.848	0.809	0.828	4.85	0.892	0.889	0.890	4.80
CMUNet¹³	0.920	0.913	0.916	3.41	0.907	0.896	0.902	3.45	0.930	0.927	0.928	3.41
NHBS¹⁴	0.873	0.890	0.882	4.53	0.886	0.872	0.879	4.55	0.922	0.916	0.919	4.52
nnUNetV2²⁴	0.915	0.902	0.910	3.60	0.896	0.891	0.881	3.65	0.919	0.923	0.920	3.60
Ours	0.932	0.913	0.922	3.31	0.905	0.907	0.906	3.35	0.919	0.945	0.931	3.30

Quick links

Search