Lumbar spine MRI annotation with IVD height and Pfirrmann grade predictions (PLOS ONE 2024)
Description
This is the data used in the paper submitted for publication in 2024 to PLOS ONE journal titled "Lumbar spine MRI annotation with IVD height and Pfirrmann grade predictions (PLOS ONE 2024)" This dataset contains 515 T2-weighted mid-sagittal slices selected from the Lumbar Spine MRI dataset (https://data.mendeley.com/datasets/k57fr854j2/2) using the image classification method (Natalia and Sudirman, 2022). Mid-sagittal images are defined as the image that is located closest to the median plane bisecting the body vertically through the midline roughly equally from the left and right side. There are six sub-folders. • DICOM – contains the 515 original DICOM images (file extension .ima) • Input_384 - contains the corresponding PNG file of each DICOM image after being scaled to 384x384 pixels • Label_384_6SG - contains the label/segmented images that were created manually. These images were used to train deep-learning image segmentation models • Inferred_384_6SG - contains the label/segmented images that were the results of the automatic segmentation using ResNet-50 architecture (the best model we found in our experiment for this dataset) • Postprocessed_384_6SG – contains the result of applying post-processing (to improve the quality of the segmentation) • IVD_HeightResultsPP – contains the automatically determined IVD mid-height data for each input image The files with filenames beginning with Hyperparameters are text files containing the result of the hyperparameter optimization process during the machine learning training. The file MLTrainingData – PP.mat is a MATLAB data file containing the extracted self-similar color correlogram feature from the last three IVD regions in each input image. References Natalia, F. and Sudirman, S. (2022) Classification of Sagittal Lumbar Spine MRI for Lumbar Spinal Stenosis Detection Using Transfer Learning of a Deep Convolutional Neural Network. Edited by A. K. Nagar et al. Singapore: Springer Nature Singapore. doi: 10.1007/978-981-16-6309-3_16.
Files
Institutions
Categories
Funding
Kementerian Riset Teknologi Dan Pendidikan Tinggi Republik Indonesia
004-RD-LPPM-UMN/ P-HD/VI/2022