Radio frequency coil failure assessment through characterization of coil noise

Motivation: Clinical imaging which occurs between coil failure and quality assurance detection of coil failure can be costly and detrimental to patient care. Efficient, automated quality assurance has the potential to reduce rescans and improve imaging reliability. Goal(s): Explore noise characteristics in clinical images acquired shortly before coil failure is detected through standard quality assurance protocols. Approach: Analyze the noise in signal obtained from well-functioning coils and near failures. We then apply machine learning techniques to detect potential failures. Results: We demonstrated the ability of the proposed approach to distinguish between images acquired from within coil failure window and images acquired from functional coils. Impact: This result paves the way for real-time quality assurance in MRI systems. The work has the potential to reduce rescans, optimize workflow efficiency, and ultimately benefit both clinical outcomes and use of healthcare resources.