Spinal-Multiple-Myeloma-SEG: Trabecular Bone Segmentation Extension

Spinal-Multiple-Myeloma-SEG: Trabecular Bone Segmentation Extension This Zenodo record contains expert-validated voxel-wise segmentation masks of the trabecular compartment of lumbar vertebral bodies in dual-energy CT examinations of patients with multiple myeloma. The release extends the publicly available Spinal-Multiple-Myeloma-SEG dataset available through The Cancer Imaging Archive (TCIA): https://doi.org/10.7937/k4qv-hh78. The dataset provides trabecular bone segmentation masks in both DICOM-SEG and NIfTI formats. Labels correspond to individual lumbar vertebrae (L1-L5 and L6 when present). The dataset contains segmentation masks from 71 of the 72 CT examinations included in the original Spinal-Multiple-Myeloma-SEG dataset. One examination was not annotated because the available CT acquisition did not include the lumbar spine region required for trabecular compartment segmentation. The annotations were generated using a pretrained nnU-Net model followed by manual correction and radiological quality control. The released masks preserve the original CT image geometry and are suitable for machine learning development, segmentation benchmarking, and quantitative vertebral bone analysis. Dataset organization, label definitions, and usage instructions are provided in the accompanying README file. Associated TCIA dataset: Spinal-Multiple-Myeloma-SEG Citation: Nohel, M., Válek V., Rohan T., Štork M., Jakubíček R., Chmelík J., Dostál M. (2026). Segmentation of spinal multiple myeloma lesions in dual-energy CT (Spinal-Multiple-Myeloma-SEG) (Version 1) [Dataset]. The Cancer Imaging Archive. https://doi.org/10.7937/k4qv-hh78