Skip to main content

Table 3 State-of-the-art methods, performance and objective

From: Survey of methods and principles in three-dimensional reconstruction from two-dimensional medical images

Model

Metric

Result

Objective

DIRNet [138]

DICE

0.80 ± 0.08

Cine MRI registration

CNN + STN [139]

SNR

207.42 ± 96.73

Tolerating respiratory motion while imaging

Unet + STN [140]

DICE

0.90 ± 0.60

Deformable medical image registration

RegNet [141]

MAE

1.19 ± 1.17

Non rigid image registration

LungRegNet [142]

TRE

1.59 ± 1.58

Unsupervised deformable image registration

SARNet [143]

DICE

0.741 ± 0.08

Data independent image registration

3D Unet + SVF [144]

DICE

0.6794 ± 0.04

Spatio-temporal regularizer

VoxelMorph [145]

DICE

0.78 ± 0.03

Deformable medical image registration

GAN + AIRnet [146]

D-scorea

0.60

Adversarial image registration

PDDNet [147]

DICE

0.584 ± 0.059

Probalistic dense displacement network

  1. aDiscrimination score