Table 2.

Experimental results on cryo-ET tomogram segmentation.

Method	mIoU	Dice
URFinder [SHREC21]	66.8	80.1
U-CLSTM [SHREC21]	77.6	87.4
MCDSNet [SHREC21]	78.4	87.9
YOPO [SHREC21]	68.1	81.0
CFN [SHREC21]	75.6	86.1
TM [SHREC21]	50.0	66.6
TM-F [SHREC21]	51.8	68.2
DeepFinder [SHREC21]	83.5	91.0
VoxResNet [NeuroImage18]	83.7±0.4	91.1±0.3
Swin UNETR [CVPR22]	79.4±0.2	88.5±0.1
MedNeXt [miccai23]	82.7±0.6	90.5±0.3
SwiFT [NeurIPS23]	81.5±0.1	89.8±0.1
MVGFormer(MLP)	84.6±0.3	91.6±0.2
MVGFormer(MF)	85.7 ±0.1	92.6 ±0.2
MVGFormer(P3DA)	86.9 ±0.2	93.1 ±0.4
MVGFormer(permutation)	86.8±0.1	93.0±0.3

We use the bold and the underlined to represent the best results and the runner-ups, respectively.