---
tags:
- image-classification
- pytorch
metrics:
- accuracy
- Cohen's Kappa

model-index:
- name: PANDA_ViT
  results:
  - task:
      name: Image Classification
      type: image-classification
    metrics:
      - name: Accuracy
        type: accuracy
        value: 0.47959184646606445
      - name: Quadratic Cohen's Kappa
        type:  Quadratic Cohen's Kappa
        value: 0.5590880513191223
---

# PANDA_ViT

An attempt to use a ViT for medical image classification (ISUP grading in prostate histopathology images). Currently uses a tiled and concatenated WSI as input

Example Images (1152,1152,3) 36 WSI patches: 



ISUP 0:
<img width="500" height="500" src="https://huggingface.co/smc/PANDA_ViT/resolve/main/0c02d3bb3a62519b31c63d0301c6843e_0.jpeg">

ISUP 1:
<img width="500" height="500" src="https://huggingface.co/smc/PANDA_ViT/resolve/main/0cee71ab57422e04f76e09ef2186fcd5_1.jpeg">
ISUP 2:
<img width="500" height="500" src="https://huggingface.co/smc/PANDA_ViT/resolve/main/00bbc1482301d16de3ff63238cfd0b34_2.jpeg">
ISUP 3:
<img width="500" height="500" src="https://huggingface.co/smc/PANDA_ViT/resolve/main/0c5c2d16c0f2e399b7be641e7e7f66d9_3.jpeg">
ISUP 4:
<img width="500" height="500" src="https://huggingface.co/smc/PANDA_ViT/resolve/main/0c88d7c7033e2048b1068e208b105270_4.jpeg">
ISUP 5:
<img width="500" height="500" src="https://huggingface.co/smc/PANDA_ViT/resolve/main/00c15b23b30a5ba061358d9641118904_5.jpeg">