Transforming MRI k-Space Into Sound: A Novel Framework for Real-Time Quality Control Using Sonification and AI

Magnetic Resonance Imaging (MRI) quality control (QC) is essential for ensuring diagnostic accuracy and reliability. Current QC methods rely heavily on image-space evaluation or specialized phantoms, which may miss subtle acquisition errors and require post hoc analysis. We propose a novel framework that transforms raw MRI k-space data into audio signals (sonification) and leverages artificial intelligence (AI) audio feature extraction techniques to detect imaging artifacts in real time. By converting k-space trajectories into audible waveforms, anomalies such as motion, ghosting, RF spikes, or gradient Instabilities manifest as distinct acoustic patterns. Using audio features such as Mel-Frequency Cepstral Coefficients (MFCCs), spectral entropy, and spectrogram representations, our system trains machine learning models to automatically identify acquisition errors. This approach complements existing image-based QC methods and offers the potential for a lightweight, audible, and real-time solution. 1