TissueMNIST

Dataset Information

We use the BBBC051, available from the Broad Bioimage Benchmark Collection. The dataset contains 236,386 human kidney cortex cells, segmented from 3 reference tissue specimens and organized into 8 categories. We split the source dataset with a ratio of 7:1:2 into training, validation and test set. Each gray-scale image is 32×32×7 pixels, where 7 denotes 7 slices. We take maximum values across the slices and resize them into 28×28 gray-scale images.

Task: multi-class

Labels:

0: Collecting Duct, Connecting Tubule, 1: Distal Convoluted Tubule, 2: Glomerular endothelial cells, 3: Interstitial endothelial cells, 4: Leukocytes, 5: Podocytes, 6: Proximal Tubule Segments, 7: Thick Ascending Limb

Samples:

• Train: 165466
• Validation: 23640
• Test: 47280

Experiment Parameter

Learning Rate: 3e-4

Training Epoch: 50

Convergence Epoch: 10