The official implementation of Trustworthy Few-Shot Transfer of Medical VLMs through Split Conformal Prediction.
📜 Medical Image Computing and Computer Assisted Intervention (MICCAI)
Julio Silva-Rodríguez, Ismail Ben Ayed, Jose Dolz ⋅ ÉTS Montréal

• Install in your environment a compatible torch version with your GPU. For example:

conda create -n scat python=3.11 -y
conda activate scat
pip install torch==2.6.0 torchvision==0.21.0 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124

git clone https://github.com/jusiro/SCA-T.git
cd FCA
pip install -r requirements.txt

• Configure data paths (see ./local_data/constants.py).
• Download, and configure datasets (see ./local_data/datasets/README.md).

We present the basic usage here.

(a) Features extraction:

• python extract_features.py --task Gleason,MESSIDOR

(b) Spit conformal prediction (SCP):

• python scp.py --task Gleason,MESSIDOR --k 16 --alpha 0.10 --ncscore lac

(c) Transductive Split Conformal Adaptation (SCA-T):

• python scat.py --task Gleason,MESSIDOR --k 16 --alpha 0.10 --ncscore lac

You will find the results upon training at ./local_data/results/.

If you find this repository useful, please consider citing this paper:

@inproceedings{scat25,
    title={Trustworthy Few-Shot Transfer of Medical VLMs through Split Conformal Prediction},
    author={Silva-Rodríguez, Julio and Ben Ayed, Ismail and Dolz, Jose},
    booktitle={Medical Image Computing and Computer Assisted Intervention (MICCAI)},
    year={2025}
}