Cardiac MRI in MINOCA: Current Evidence, Parametric Mapping Advances, and Future AI Applications-A Systematic Review

Background: Myocardial infarction with nonobstructive coronary arteries (MINOCA) is a heterogenous clinical syndrome in which aetiologies range from "true" ischemic mechanisms to non-ischemic mimics (e.g., myocarditis and Takotsubo syndrome). Cardiac magnetic resonance (CMR) plays a central role in the diagnostic pathway. Recent advances in parametric mapping (native T1, T2, and extracellular volume ECV) and evolving AI/radiomic methods promise to further improve diagnostic accuracy and prognostic stratification. This review aims to evaluate the current CMR evidence in MINOCA, while highlighting parametric mapping advances and future directions in the sphere of AI and radiomics. Methods: A systematic literature search of PubMed and the Directory of Open Access Journals (DOAJ) was performed. We included original prospective and retrospective CMR studies of MINOCA and MINOCA-like presentations in adults. Data were extracted into a master dataset and synthetised thematically into five subsections: (1) diagnostic yield, (2) reclassification rate), (3) timing of CMR, (4) prognosis, and (5) future directions. Results: Twenty-two studies met the inclusion criteria. CMR diagnostic yield varied by protocol and timing but was consistently substantial. CMR consistently reclassified initial MINOCA diagnoses (ischemia or alternative non-ischemic diagnoses). Parametric mapping provided incremental diagnostic and prognostic information. Across studies, early imaging (ideally within the first 1-2 weeks) increased diagnostic yield, while delayed CMR reduced detectability of transient lesions. Early AI and radiomics work show promise for LGE-based classification and for predicting post-contrast findings from non-contrast data, but current models require larger, multicentre training and robust external validation. Conclusions: CMR increases diagnostic yield and reclassification rates in MINOCA, particularly when performed early and with standardised T1/T2/ECV mapping. Mapping not only improves detection of inflammatory and diffuse injuries but also contributes to prognostic stratification. High-resolution LGE, OCT, and AI/radiomics are promising future refinements but need prospective validation in large, early, mapping-inclusive cohorts.