Service Lines
Digital Health
Technology
Service Line Focus: Research Shows AI’s Potential in Spine Care
AI applied to spine imaging shows potential to improve diagnostic accuracy and efficiency (eg, radiologist workflow), bridge capacity gaps, and elevate patient care. Two novel research concepts in this arena highlighted at the 2024 Radiological Society of North America (RSNA) conference include:
- DLM standardizing image interpretation. Lumbar spine MRI remains the preferred imaging modality for diagnosing chronic low back pain. However, interpretation can be subject to inter-reader variability. A poster study from Harvard Medical School highlighted a deep learning model (DLM) that reviewed nine degenerative spinal conditions across a data set of 55,000 lumbar spine MRI studies affiliated with Mass General Brigham. Researchers found the model was able to achieve accuracy rates of 77% to 97% across all pathologies of interest, showing potential in standardizing image interpretation.
- Workflow and AI prioritization. Researchers from the University of Washington, Seattle, studied the impact of an AI prioritization model on radiologist workflow for detecting cervical spine fractures in the outpatient setting. They focused specifically on the wait time between CT scan completion and a radiologist initiating the case for review. The study analyzed 2,009 CT scans and found the scans that leveraged the AI model reduced wait time 56% vs scans that did not include the AI model (99 minutes vs 225.7 minutes). Overall, the AI model demonstrated positive benefits on radiologist workflow by prioritizing scans and accelerating time to treatment.
While AI adoption is still nascent, these advancements represent a strategic opportunity for health care programs to deliver more efficient, accurate and consistent patient care. Spine programs should consider both clinical and nonclinical use cases as well as create a framework to measure key value metrics (eg, financial, clinical, operational).
Sg2 offers members in-depth strategic guidance in numerous service lines: behavioral health, cancer, cardiovascular, medicine, neurosciences, orthopedics, surgery and women’s health. Not a member? Reach out to us at learnmore@sg2.com for information on the expert intelligence, data-driven insights and strategic perspective Sg2 offers to health systems nationwide.
Sources: Wu K. DeepSpine: a comprehensive deep learning model for multi-task lumbar spine MRI analysis. Presented at RSNA 2024, Chicago, December 4, 2024; Mahdavi A. Assessing the impact of an AI-assisted prioritization system on cervical spine fracture detection through CT imaging in the outpatient setting. Presented at RSNA 2024, Chicago, December 4, 2024; Yee KM. AI reduces time to interpretation of CT cervical spine exams. AuntMinnie.com. December 4, 2024.