CardioSYNTAX dataset

This is the accompanying dataset for the paper accepted to the WACV 2025 conference, titled:

"CardioSyntax: end-to-end SYNTAX score prediction – dataset, benchmark and method"

The data is structured as follows: the root directory contains both metadata and data directories. The all.json file includes metadata for all parts of the dataset, while the partX.json files only contain metadata for the corresponding part of the dataset.

The partX.zip files include angiographic videos in the NumPy npy format, grouped by study.

Additionally, the file three_experts.json contains a subset of 60 samples that were annotated by three experts, with expert 0 being the same as the whole dataset.