Whole-Body Bone Scintigraphy Segmentation Masks, Ground-Truth Annotations for Bone Area Segmentation

Ground-truth annotation masks for 3,593 whole-body bone scintigraphy images from the BS-80K dataset (Huang et al., 2022). Covers anterior (1,841 images) and posterior (1,752 images) views. Twelve skeletal regions annotated: skull, cervical vertebrae, thoracic vertebrae, ribs, sternum (anterior only), clavicle, scapula, humerus, lumbar vertebrae, sacrum, pelvis, and femur.

Masks are provided as grayscale PNGs (pixel values 0–12, label indices) and RGB-colored PNGs for visualization. Annotation was performed using IbisPaintX, supervised by nuclear medicine physicians at the Department of Nuclear Medicine and Molecular Theranostics, Dr. Hasan Sadikin General Hospital, Faculty of Medicine, Universitas Padjadjaran, Indonesia.

Associated code and split files: https://github.com/Liamours/wbbs-nnunetv2

This dataset accompanies the paper: "Deep Learning-Based Segmentation of Whole-Body Bone Scan Images Using nnU-Netv2", International Journal of Intelligent Engineering and Systems (IJIES-INASS).