README.md

End-to-end SYNTAX score prediction: benchmark and methods

Run inference:

python seq_apply.py -r DATASET_DIR -w WEIGHTS_DIR --fold 0

DATASET_DIR - dataset directory.
Sample dataset: https://disk.yandex.com/d/drZKKBJnH2r8vg

WEIGHTS_DIR - model weights directory.
Sample weights: https://disk.yandex.com/d/_4ARTacETFQr1A

Train backbone:

mkdir backbone
# train backbone model for left coronary artery
python backbone_train.py -r DATASET_DIR -a left --fold 0
# train backbone model for right coronary artery
python backbone_train.py -r DATASET_DIR -a right --fold 0

Train study-level sequence model

mkdir seq_models
# train sequence model for left coronary artery
python seq_train.py -r DATASET_DIR -a left --fold 0 --variant lstm_mean
# train sequence model for right coronary artery
python seq_train.py -r DATASET_DIR -a right --fold 0 --variant lstm_mean