Renal MRI standardisation and clinical validation to improve personalised management of patients with chronic kidney disease

Renal magnetic resonance imaging (MRI) is a powerful non-invasive technique that enables detailed characterisation of kidney structure, microstructure, and function, with no need for ionising radiation or contrast agents. Chronic kidney disease (CKD) is a growing global health challenge. Non-invasive biomarkers to early detect and monitor CKD are needed, and multiparametric MRI shows strong potential to answer this need. Yet renal MRI broader clinical translation remains limited by the lack of standardised acquisition and analysis protocols and by insufficient clinical validation. This PhD project aimed at bridging these gaps by advancing renal MRI technical standardisation and providing evidence of its clinical validity. Within the multinational RESPECT project, dedicated MRI phantoms were employed across multiple centres and vendors to harmonise T1, T2, and diffusion-weighted imaging (DWI) protocols, providing the basis for quality control procedures and reproducibility benchmarks. Standardised renal MRI was repeatedly acquired in healthy volunteers (HVs) from one centre, demonstrating high repeatability and reproducibility for both renal blood flow (RBF) measurements from phase-contrast (PC) MRI and DWI-based microstructural parameters. With a view to standardise kidney segmentation, besides MRI acquisition, available artificial intelligence (AI)-based segmentation methods were compared with the gold standard manual planimetry on anatomical renal MRI scans from both HVs and patients with autosomal dominant polycystic kidney disease (ADPKD). Most AI-based tools showed promising accuracy, while substantially reducing the segmentation time. In addition, a novel DWI-based approach was developed to quantify cyst burden and enable separate characterisation of cystic and non-cystic renal tissue in ADPKD. Multiparametric renal MRI was then applied to patients with different kidney diseases, in particular ADPKD and complement-related glomerulonephritis, highlighting its potential for clinical translation as a non-invasive tool able to capture microstructural and functional changes and thus to monitor kidney disease progression and response to therapy, towards personalised patient management.