Synthetic Cerebral Digital Subtraction Angiography (DSA) via Deep Generative Modelling: Images for Reader Study

Synthetic Cerebral DSA Dataset This repository contains 400 synthetic cerebral digital subtraction angiography (DSA) images of the anterior and posterior circulations. The images were generated using a diffusion model and were used in a reader study. Dataset Structure The dataset contains 400 images in total, organized as: 10 runs × 4 angiographic views × 10 synthetic images = 400 images synthetic_DSA/ └── <runID>/ # 10 runs: run1, ..., run10 └── img/ ├── anterior_single_a_0_0/ # 10 images ├── anterior_single_b_-90_0/ # 10 images ├── posterior_single_a_0_0/ # 10 images └── posterior_single_b_-90_0/ # 10 images Naming Convention Subfolders are named according to the following structure: {circulation}_{plane}_{primary_angle}_{secondary_angle} Subfolder Circulation Plane Primary Angle Secondary Angle anterior_single_a_0_0 Anterior Single A 0° 0° anterior_single_b_-90_0 Anterior Single B -90° 0° posterior_single_a_0_0 Posterior Single A 0° 0° posterior_single_b_-90_0 Posterior Single B -90° 0° The two angles correspond to the DSA angiography C-arm (X-ray tube / detector) angles: primary angle and secondary angle. Files Images are provided as PNG files and are named using a zero-padded index: sample_XXXXXX.png Data Note All images in this dataset are synthetic and were generated by a diffusion model. The dataset does not contain original patient DSA images.