Table 1 Summarization of downstream task datasets

From: Adaptive compression framework for giga-pixel whole slide images

Data level

Dataset

Organ

Sample count

Patch size

Magnification

Task

Patch-level

NCT-CRC

Colon

107,180

224 × 224

20×

Cls (9)

 

MHIST

Colon

3152

224 × 224

40×

Cls (2)

 

LI

Stomach

973

512 × 512

Cls (2)

 

SICAPv2

Prostate

18,786

512 × 512

10×

Cls (4)

 

WSSS4LUAD

Lung

4695

Dynamica

40×

Cls (2)

 

SNOW

Breast

5000b

512 × 512

40×

Seg (2)

 

NuInsSeg

Various

474

512 × 512

40×

Seg (2)

 

PanNuke

Various

2658

256 × 256

40×

Seg (2)

Side-level

TCGA-RCC

Kidney

884

512 × 512

20×

Cls (3)

 

TCGA-NSCLC

Lung

993

512 × 512

20×

Cls (2)

 

TCGA-BRCA

Breast

1008

512 × 512

20×

Cls (2)

 

Camelyon16

Breast

268

512 × 512

40×

Cls (2)

 

CBTN

Brainc

300

512 × 512

20×

Cls (2)

  1. The number in the bracket indicates the number of classes.
  2. Cls classification, Seg segmentation.
  3. aThe width and height range from 150 to 300, with an average size of 220 × 222.
  4. bThe original SNOW dataset has 20k images. We sampled 5k images.
  5. cPediatric tissue samples.