Lumbar spine segmentation in MR images: a dataset and a public benchmark.

Autor: van der Graaf JW; Diagnostic Image Analysis Group, Radboud University Medical Center, Nijmegen, The Netherlands. jasper.vandergraaf@radboudumc.nl.; Department of Orthopedic surgery, Radboud University Medical Center, Nijmegen, The Netherlands. jasper.vandergraaf@radboudumc.nl., van Hooff ML; Department of Orthopedic surgery, Radboud University Medical Center, Nijmegen, The Netherlands.; Department Research, Sint Maartenskliniek, Nijmegen, The Netherlands., Buckens CFM; Department of Medical Imaging, Radboud University Medical Center, Nijmegen, The Netherlands., Rutten M; Diagnostic Image Analysis Group, Radboud University Medical Center, Nijmegen, The Netherlands.; Department of Radiology, Jeroen Bosch Hospital, 's-Hertogenbosch, The Netherlands., van Susante JLC; Department of Orthopedic Surgery, Rijnstate Hospital, Arnhem, the Netherlands., Kroeze RJ; Department of Orthopedic Surgery, Sint Maartenskliniek, Nijmegen, The Netherlands., de Kleuver M; Department of Orthopedic surgery, Radboud University Medical Center, Nijmegen, The Netherlands., van Ginneken B; Diagnostic Image Analysis Group, Radboud University Medical Center, Nijmegen, The Netherlands., Lessmann N; Diagnostic Image Analysis Group, Radboud University Medical Center, Nijmegen, The Netherlands.
Jazyk: angličtina
Zdroj: Scientific data [Sci Data] 2024 Mar 02; Vol. 11 (1), pp. 264. Date of Electronic Publication: 2024 Mar 02.
DOI: 10.1038/s41597-024-03090-w
Abstrakt: This paper presents a large publicly available multi-center lumbar spine magnetic resonance imaging (MRI) dataset with reference segmentations of vertebrae, intervertebral discs (IVDs), and spinal canal. The dataset includes 447 sagittal T1 and T2 MRI series from 218 patients with a history of low back pain and was collected from four different hospitals. An iterative data annotation approach was used by training a segmentation algorithm on a small part of the dataset, enabling semi-automatic segmentation of the remaining images. The algorithm provided an initial segmentation, which was subsequently reviewed, manually corrected, and added to the training data. We provide reference performance values for this baseline algorithm and nnU-Net, which performed comparably. Performance values were computed on a sequestered set of 39 studies with 97 series, which were additionally used to set up a continuous segmentation challenge that allows for a fair comparison of different segmentation algorithms. This study may encourage wider collaboration in the field of spine segmentation and improve the diagnostic value of lumbar spine MRI.
(© 2024. The Author(s).)
Databáze: MEDLINE
Externí odkaz: