Dataset for the Segmentation of Cervical Spinal Cord and Cervical MS Lesions
Description
The cervical spinal cord is a segment of the spinal cord located in the neck of the spine and forms the transition zone for many important functional pathways in the body. The cervical region is located approximately at the level of the C1-C7 intervertebral disc of the spinal cord. Accurate segmentation of the cervical spinal cord is therefore a critical step in the diagnosis and treatment of neurological disorders. This dataset was prepared for the detection and segmentation of cervical spinal cord cross-sectional area (CSA) and MS (multiple sclerosis) lesions in the cervical spinal cord. The MR images in the dataset are T2-weighted (T2-w) modality in axial plane and were collected from volunteer patients at Akdeniz University Hospital. The preparation of the dataset was conducted as a retrospective study with the relevant decision of the Clinical Research Ethics Committee of the Faculty of Medicine of Akdeniz University . The images in the dataset were obtained from a total of 87 patients diagnosed with MS in the last 2 years until May 2024, without personal information of the patients. Of these patients, 68 were female and the remaining 19 were male, ranging in age from 14 to 72 years. MR scans of the patient group were performed on a Siemens Magnetom Spectra 3.0T MR scanner. Data were collected using 2D MR slices in T2-w modality in a turbo spin echo sequence scanned in the axial (transverse) plane of the cervical spinal cord region, which is requested from MS patients as part of the clinical routine. The slice thickness for MR slices in the scans in the dataset was 4 mm. The MR slices in the dataset were initially provided in DICOM format and were converted to NIfTI format during the pre-processing steps. Subsequently, from these MR scans, 231 suitable slices were obtained in .png format. ITK-Snap software was used for this data conversion. The MR scans in the dataset were reviewed by two different radiologists with consensus from Akdeniz University, and the ground truth masks were performed manually in ITK-Snap, and the selected slices were double-checked and verified. In this study, two different labels were created by experts on MR slices. The first label indicates the area of the MS lesion in the CSA region (grey area), while the other label identifies the remaining area (white area), CSA without MS, after subtracting the MS lesion from the CSA. To share the dataset of 231 MR images, two folders were created here: images and ground truth masks. The images folder contains MR slices in .png format, while the ground truth masks folder contains ground truth masks delineated by radiologists for MR slices in .png. In the folders, the first patient data is patient_0 and the last patient is patient_86. Multiple slices from the same patient can be present in the dataset and are separated by slice number. In the ground truth masks, the white area represents the CSA region, while the gray area represents the MS lesion in the CSA region.
Files
Steps to reproduce
You can send your questions, comments, suggestions and reproduction notices about the dataset to Dr. Emre Dandıl (emre.dandil@bilecik.edu.tr).
Institutions
Categories
Funding
Bilecik University
2021- 01.BŞEÜ.03-02