ASL-MRI quantification involves kidney segmentation and cortex-medulla differentiation to obtain cortical renal blood flow, requiring time consuming manual interaction hampering clinical adoption. We applied machine learning to automat renal ASL-MRI quantification. A cascade of three U-nets was constructed to replace manual segmentation steps. Automatic segmentation yielded a dice score of 0.78, which was similar to the inter-observer variability of 0.77. Moreover, good agreement for cortical RBF was found between automatic and manual segmentations on group and individual level; 211±31 and 208±31mL/min/100g, respectively. Our proposed method automates quantification without compromising performance. This makes renal ASL-MRI more attractive for clinical application.
This abstract and the presentation materials are available to members only; a login is required.