The Dresden Surgical Anatomy Dataset

The Dresden Surgical Anatomy Dataset is a comprehensive medical imaging dataset designed for semantic segmentation of abdominal organs and vascular structures in laparoscopic surgery. It provides pixel-wise annotated images of key anatomical structures, making it an essential resource for researchers in computer vision, AI-driven medical imaging, and robotic surgery.

Key Points

• Dataset Type: High-resolution semantic segmentation dataset for abdominal organs and vascular structures
• Use Case: AI-powered medical imaging, robotic-assisted surgery, laparoscopic training, computer vision research
• Number of Surgeries: 32 real-world laparoscopic surgeries
• Included Anatomical Structures:
  • 8 abdominal organs: Colon, Liver, Pancreas, Small Intestine, Spleen, Stomach, Ureter, Vesicular Glands
  • Abdominal wall
  • 2 vascular structures: Inferior Mesenteric Artery, Intestinal Veins
• Dataset Size: At least 1,000 fully annotated images per organ/structure
• Video Source: Da Vinci® Xi/X Endoscope (8mm, 30° angled camera)
• Resolution & Format: 1920 × 1080 pixels, MPEG-4
• Annotation Tools Used:
  • 3D Slicer 4.11 (for precise pixel-wise segmentation)
  • Surgery Workflow Toolbox [Annotate] v2.2.0
• Weak Labels Included: Visibility annotations for anatomical structures in each image
• Diverse Patient Pool:
  • 26 male patients out of 32
  • Average age: 63 years
  • Mean BMI: 26.75 kg/m²
• Ideal For:
  • Medical AI & Deep Learning Research
  • Surgical Navigation & Augmented Reality (AR) Applications
  • Laparoscopic & Robotic-Assisted Surgery Training
  • Biomedical Image Processing & Semantic Segmentation Studies