Potential Impacts of Artificial Intelligence on Spine Imaging Interpretation and Diagnosis
Spine and related disorders represent one of the most common causes of pain and disability in the United States. Imaging represents an important diagnostic procedure in spine care. Imaging studies contain actionable data and insights undetectable through routine visual analysis. Convergent advances in imaging, artificial intelligence (AI), and radiomic methods has revealed the potential of multiscale in vivo interrogation to improve the assessment and monitoring of pathology. AI offers various types of decision support through the analysis of structured and unstructured data. The primary purpose of this qualitative exploratory case study was to identify the potential impacts of AI solutions on spine imaging interpretation and diagnosis. Selected constructs from the diffusion of innovations theory and the technology acceptance model provided the conceptual framework. Data were acquired from 4 consensus-based white papers, researcher reflective journaling, and 2 homogenous focus group sessions comprising radiologists and AI experts. Content and thematic analyses of acquired data were performed with ATLAS.ti. Three primary themes emerged from qualitative analysis: patient-based decision support, population-based decision support, and application-based decision support. Subthemes include multiscale in vivo analysis, naturally language processing, change analysis, prioritization, and ground truth. The results suggest how further development of AI could fundamentally alter how spine pathology is detected, characterized, and classified. The study also addresses the potential impact of AI on in vivo tissue analysis, the differential diagnosis, and imaging workflow. This includes introducing the concept of the virtual biopsy and its use in spine imaging.