Medical Lumbar Spine 3D Axial MRI Dataset for Stenosis Detection, Severity Classification, and Anatomical Segmentation
Description
☐ Dataset Discription: ● The Medical Lumbar Spine 3D Axial MRI Dataset contains 500 anonymized axial lumbar spine MRI examinations from 500 patients acquired at Fırat University Hospital (FUH) under the bilateral protocol of join international research project between Korea and Türkiye. MRI scans were obtained using routine T2-weighted axial sequences on Philips Ingenia (1.5T/3T), GE Signa HDxt (1.5T), and GE Signa Excite (1.5T) scanners. The dataset supports research on lumbar spinal stenosis localization, severity classification, and axial-wise anatomical segmentation. ● The dataset consists of three components: 1. 3D DICOM Volumes: Original 3D axial MRI volumes organized by patient, covering lumbar levels starting L1–L2 to L5–S1 (5 vertebrae). All DICOM files were anonymized to remove patient-identifying information while preserving essential imaging metadata. 2. Localization and Severity Classification: Axial PNG images with Pascal VOC XML annotations. Expert-labeled bounding boxes identify five anatomical regions: Central Canal Stenosis (CCS), Left and Right Lateral Recess Stenosis (LLS and RLS), and Left and Right Foraminal Stenosis (LFS and RFS). Each annotation includes bounding-box coordinates, lumbar level (L1-L2 to L5-S1), AP diameter (mm), anatomical region, and stenosis grade (Normal, Stenosis, or Severe Stenosis). 3. Anatomical Segmentation: Axial MRI slices with pixel-wise masks for Posterior Elements (PE), Central Canal (CC), Area of Anterior–Posterior (AAP), and Intervertebral Disc (IVD). Mask labels are encoded for visualization as: 0 (Background), 50(PE), 100(CC), 150(AAP), and 200(IVD). Masks were generated using AI and refined through manual correction and expert neuroradiologists' validation. ☐ ACKNOWLEDGMENT: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. RS-2023-00256517) and by the TUBITAK (The Scientific and Technological research Council of Turkey) under Grant Number: 123N325. ☐ Please cite these articles when you use our dataset: [1] Abdulmahmod, Osamah F., Mugahed A. Al-antari, Hyunwook Kwon, Afnan Habib, Mukhlis Raza, Metin Kaplan, Bilal Ertuğrul, İsmail Akçin, Ertan Bütün, and Yeong Hyeon Gu. "Medical Spine Sagittal MRI Dataset for Segmentation and Foraminal Stenosis Detection." Scientific Data 13 (2026): Article 809. https://doi.org/10.1038/s41597-026-07138-x [2] Al-Antari, Mugahed A., Saied Salem, Mukhlis Raza, Ahmed S. Elbadawy, Ertan Bütün, Ahmet Arif Aydin, Murat Aydoğan, Bilal Ertuğrul, Muhammed Talo, and Yeong Hyeon Gu. "Evaluating AI-powered predictive solutions for MRI in lumbar spinal stenosis: a systematic review." Artificial Intelligence Review 58, no. 8 (2025): 221.
Files
Steps to reproduce
• The Medical Lumbar Spine 3D Axial MRI Dataset is well organized and easy to use. To reproduce and utilize the dataset: 1. Download the compressed ZIP package from Mendeley Data. 2. Extract the archive to access the DICOM, Localization_and_Severity_Classification, and Segmentation folders. 3. Load the DICOM volumes using your preferred medical imaging software or Python libraries. 4. Match each image with its corresponding localization, severity classification, or segmentation annotations. 5. Train, validate, and evaluate AI models for lumbar spine analysis using the provided annotations.
Institutions
- Sejong UniversitySeoul, Seoul
- Fırat UniversityElazığ, Elâzığ