Incidence of CNS infarcts remains a serious complication of surgical aortic valve replacement (SAVR). Although most CNS infarcts do not lead to clinical stroke, they are associated with neurological complications such as post-operative delirium (PoD). Identifying image-based brain biomarkers associated with PoD is an important step towards identifying individuals with high PoD risk. We analyzed MRI-derived lesionality and connectome data, as well as cognitive outcomes, from SAVR patients (n=308). This data was originally collected as part of the clinical trial ‘Neuroprotection in Patients Undergoing Aortic Valve Replacement’ (NCT02389894, PMID:28787505). Due to the zero-inflated nature of the data, we utilized a zero-inflated gaussian modeling approach (PMID:33166356), originally developed for micro-biome analysis, to identify region-specific peri-operative increases in lesionality/connectivity that were associated with developing PoD, most notably in Cerebellar regions. This analysis paves the way towards identifying MRI-derived image-based biomarkers for peri-operative neurological complications and highlights the utility of zero-inflated modeling approaches.