A Semi-automated Method to Measure the Height of Lumbar Intervertebral Discs

Measuring the height of a lumbar intervertebral disc is an important step in categorizing the degenerative level of the disc. Popular disc categorization techniques, such as the Pfirrmann grading system, use several visual features that physicians can observe directly from the medical images – including the disc height. In this paper, we propose a semiautomated method for measuring the average height of a lumbar intervertebral disc from the sagittal Magnetic Resonance Images of lumbar vertebrae. The technique uses a combination of approaches starting with selecting a mid-sagittal slice from each patient study using a deep learning model, manual image segmentation, linear regression to find a mid-height line, and geometry to split the disc region into its upper and lower regions and calculate its height. We show that our results generally compare well with others and identify an area that warrants further investigation.