2024 ARRS ANNUAL MEETING - ABSTRACTS

RETURN TO ABSTRACT LISTING


E5524. The Evolution of Lumbar Spine Stenosis Evaluation: From Grading Systems to Artificial Intelligence
Authors
  1. Peter Pham; University of California Irvine
  2. Joseph Burns; University of California Irvine
  3. Maryam Golshan-Momeni; University of California Irvine
  4. Arash Anavim; University of California Irvine
  5. Lawrence Wang; University of California Irvine
Background
Degenerative stenosis of the lumbar spine is a leading cause of dysfunction and pain in middle age and older adults, typically reported as back and/or leg pain, progressive with standing or walking. Deficiency of clear-cut imaging criteria and parameters to uniformly characterize lumbar spine stenosis may lead to subjective radiology interpretation. As such, multiple grading systems have evolved to standardize imaging interpretation of lumbar spine stenosis. Development of a standardized grading system will aid in patient management, optimization of treatment outcomes, and development of research studies to evaluate treatment efficacy. Specifically, standardized imaging criteria can have a great impact in treatment consideration for conservative versus surgical management, and in cases of surgical management, can have implications in the surgical procedure and approach. This educational exhibit will review the evolution of common spinal canal stenosis grading systems and discuss a recent deep learning model that may add more value to radiology interpretations.

Educational Goals / Teaching Points
This exhibit contains a brief pictorial review of relevant anatomy, mechanisms, and nomenclature of degenerative lumbar stenosis. The evolution of popular classification systems for spinal stenosis from quantitative measurements to morphologic grading will be reviewed, specifically the Schizas, Lee, and Miskin-Mandell grading systems. The strengths and limitations of each will be highlighted. Future directions for degenerative lumbar stenosis interpretation will also be discussed, specifically the role of artificial intelligence/deep learning.

Key Anatomic/Physiologic Issues and Imaging Findings/Techniques
This review of each MRI lumbar stenosis classification system will highlight the anatomic components/landmarks considered relevant in determining the varying degrees of stenosis, such as dural sac morphology, CSF space, nerve root space, lateral recess, and facet joints. Examples will be presented using images from literature and our institution. Images generated by a deep learning model will be shown to demonstrate consistency with manual grading systems.

Conclusion
Imaging interpretation of degenerative lumbar stenosis can be subjective and widely variable. It is important for radiologists to understand different classifications used, as radiologists and surgeons may have their own preferred classification. Standard reporting by radiologists can help surgeons in their treatment decisions and surgical management. Knowledge of the strengths and limitations of each classification can help with development of novel spinal stenosis evaluation methods that employ artificial intelligence/deep learning. New advanced deep learning models may add extra value to radiologist interpretations via speed, quality, standardization, and potentially improved outcomes.