Table 1 Generation performance on five publicly available datasets evaluated by MAE, PSNR, MI, and SSIM. The bold entries in this table indicate the algorithm which gets the best performance in each task. The standard for choosing the best algorithm is to have statistical significance over the other algorithms (p-value < 0.05). If an algorithm gets the best evaluation metrics but has no statistical significance over the others (p-value > 0.05), all of them will be regarded as the best algorithms. The result show that our IMT approach outperforms both Random Forest (RF) based method⁵ and Context-Aware GAN (CA-GAN)³⁰ method on most datasets.

BraTs2015

T1  →  T2

6.025

24.717

0.617

0.910

11.947

19.738

0.787

0.826

8.292

22.560

0.862

0.866

10.692

20.301

0.788

0.575

8.654

22.517

0.901

0.880

T2  →  T1

7.921

23.385

0.589

0.893

16.587

17.462

0.661

0.723

9.937

22.518

0.777

0.854

15.430

18.507

0.673

0.723

10.457

22.374

0.818

0.896

T1  →  T2-Flair

8.176

23.222

0.609

0.873

13.999

19.157

0.722

0.756

7.934

22.687

0.833

0.837

11.671

19.969

0.749

0.797

8.462

22.642

0.879

0.857

T2  →  T2-Flair

7.318

23.138

0.610

0.875

12.658

18.848

0.756

0.749

8.858

21.664

0.848

0.836

10.469

20.656

0.817

0.823

8.950

21.791

0.928

0.860

Iseg2017

T1  →  T2

3.955

28.028

0.803

0.902

12.175

21.992

0.804

0.690

3.309

29.979

0.931

0.887

8.028

22.860

0.782

0.748

3.860

28.874

0.993

0.913

T2  →  T1

11.466

22.342

0.788

0.808

17.151

18.401

0.789

0.662

9.586

23.610

0.868

0.745

17.311

18.121

0.777

0.620

10.591

23.325

0.880

0.754

MRBrain13

T1  →  T2-Flair

7.609

24.780

1.123

0.863

13.643

19.503

0.805

0.782

6.064

26.495

1.066

0.823

9.906

22.616

1.009

0.785

6.505

26.299

1.185

0.881

ADNI

PD  →  T2

9.485

24.006

1.452

0.819

16.575

19.008

0.674

0.728

6.757

26.477

1.266

0.812

7.211

26.330

1.184

0.779

4.898

29.089

1.484

0.891

T2  →  PD

5.856

29.118

1.515

0.880

17.648

18.715

0.659

0.713

4.590

31.014

1.381

0.856

5.336

29.032

1.282

0.820

5.055

30.614

1.536

0.881

RIRE

T1  →  T2

38.047

12.862

0.694

0.501

18.625

18.248

0.724

0.749

5.250

28.994

0.636

0.736

13.690

21.038

0.513

0.506

9.105

28.951

0.698

0.760

T2  →  T1

17.022

19.811

0.944

0.622

23.374

16.029

0.650

0.728

9.035

24.043

0.916

0.692

13.964

20.450

0.737

0.538

9.105

24.003

0.969

0.741