Datasets
Standard Dataset
Test Dataset for InfoBot-GAN paper
- Citation Author(s):
- Submitted by:
- Cong Liu
- Last updated:
- Fri, 03/22/2019 - 10:43
- DOI:
- 10.21227/4vm4-bq10
- License:
113 Views- Categories:
Abstract
The dataset contains 208 patient scans that spread over three parts of anatomy(head, neck, and pelvis). The dataset aims to establish accurate anatomical correspondences between MegaVoltage Planar Digital Radio-graphs (MV-DRs) andKiloVoltage Digital Reconstructed Radiographs (KV-DRRs), which are widely used in Image-Guided Radiation Therapy (IGRT) to verify patients’ positions for accurate radiotherapy delivery.
Each scan involves eight images: the anteroposterior view KV-DRR (filename: UID_0_bone_DRR.png), the anteroposterior view KV-DRR annotation(filename: UID_0_bone_DRR_Lines.png), the anteroposterior view MV-DR (filename: UID_0_DR.png), the anteroposterior view MV-DR annotation(filename: UID_0_DR_Lines.png), the lateral view KV-DRR (filename: UID_90_bone_DRR.png), the lateral view KV-DRR annotation(filename: UID_90_bone_DRR_Lines.png), the lateral view MV-DR (filename: UID_90_DR.png), the lateral view MV-DR annotation(filename: UID_90_DR_Lines.png). All image pairs are manually aligned by the clinicians, and large appearance variations and misleading correspondences present in the pairs. The contours of bone anatomy structures are annotated by the clinicians. Images have a resolution of 0.5mm per pixel.
