AutoGlom: software tool for segmentation and analysis of magnetic resonance images of the kidney

Magnetic resonance imaging (MRI) is increasingly important in preclinical and clinical investigations of the kidney. However, there are few user-friendly, flexible, and standardized tools for evaluating MR images for quantitative imaging analysis. Here, we develop AutoGlom, an open-source, modular, and expandable imaging software tool that incorporates artificial intelligence (AI) for segmentation, analysis, and visualization of three-dimensional (3-D) MR images of the kidney. This initial version of AutoGlom focuses on morphological segmentation and quantification. We describe kidney segmentation from MR images, followed by the use of the graphical user interface of AutoGlom. Using AutoGlom, we measure glomerular number and volume from ex vivo cationic ferritin-enhanced MRI (CFE-MRI) in mice. We further demonstrate a 3-D-printed holder to allow for simultaneous imaging of up to 16 mouse kidneys at high resolution (50 μm) within several hours. The streamlined workflow facilitates rapid image analysis and accelerates optimization of cationic ferritin dosing and imaging parameters. These tools are a resource for the kidney community that may accelerate the identification of candidate imaging biomarkers from 3-D MRI of the kidney and have the potential to be extended to in vivo studies and other imaging modalities.<b>NEW & NOTEWORTHY</b> We present AutoGlom, an open-source software for quantitative kidney MRI analysis. AutoGlom integrates deep learning-based glomerular segmentation, parameter tuning, and visualization within a user-friendly interface. It enables high-throughput analysis using a 3-D-printed holder for simultaneous imaging of multiple kidneys and introduces a new image quality metric, glomerular contrast, to improve reliability. AutoGlom provides standardized, reproducible workflows for glomerular quantification, bridging preclinical and translational kidney imaging and enabling future physiological discoveries.