metadata

license: cc-by-nc-sa-4.0

RetCCL-based ABMIL models for metastasis detection

These are weakly-supervised, attention-based multiple instance learning models for binary metastasis detection (normal versus metastasis). The models were trained on the CAMELYON16 dataset using RetCCL embeddings.

Data

Training set consisted of 243 whole slide images (WSIs).
- 143 negative
- 100 positive
  - 52 macrometastases
  - 48 micrometastases
Validation set consisted of 27 WSIs.
- 16 negative
- 11 positive
  - 6 macrometastases
  - 5 micrometastases
Test set consisted of 129 WSIs.
- 80 negative
- 49 positive
  - 22 macrometastases
  - 27 micrometastases

Evaluation

Below are the classification results on the test set.

Seed	Sensitivity	Specificity	BA	Precision	F1
0	0.612	0.838	0.725	0.698	0.652
1	0.531	0.950	0.740	0.867	0.658
2	0.551	0.950	0.751	0.871	0.675
3	0.612	0.925	0.769	0.833	0.706
4	0.531	0.950	0.740	0.867	0.658

How to reuse the model

The model expects 128 x 128 micrometer patches, embedded with the RetCCL model.

import torch
from abmil import AttentionMILModel

model = AttentionMILModel(in_features=2048, L=512, D=384, num_classes=2, gated_attention=True)
model.eval()
state_dict = torch.load("seed3/model_best.pt", map_location="cpu", weights_only=True)
model.load_state_dict(state_dict)

# Load a bag of features
bag = torch.ones(1000, 2048)
with torch.inference_mode():
    logits, attention = model(bag)