Pre-processing Spine MRI to Expedite Disc Segmentation

• Makana Steinmetz, Biomedical Engineering, University of Delaware

• Dawn Elliott, Biomedical Engineering, University of Delaware

Low back pain is a widespread issue that requires an accurate understanding of intervertebral disc mechanics for enhancing diagnosis and treatment. This project aims to evaluate in vivo disc mechanics with MRI under various loading scenarios and establish a disc function baseline in healthy subjects. To analyze the MRI images effectively, segmentations of the discs are required. Previously, manual segmentation required about 2 hours per spine, making it very time-consuming and prone to human error. To address these limitations, a code was developed to automate the segmentation process which reduced human error and processing time. Before images can be processed, the images must be resized, contrast enhanced, and image type changed. A Python-based code was developed in order to expedite these preprocessing steps. The code takes the raw MRI images as inputs and outputs an adjusted tiff file for further processing steps. The preprocessing code demonstrated successful conversion of the input image to a 512 x 512 pixel arrangement, contrast enhanced, 8-bit image ready for auto segmentation. By utilizing this preprocessing code, images are ready for auto segmentation in ~15 seconds, as opposed to manual preprocessing which took ~5 minutes per spine. This streamlined approach will expedite segmentation for further analysis, including evaluation of mechanical behavior and trends in the spine with subject age.