Automated renal segmentation in healthy and chronic kidney disease subjects using a convolutional neural network.

PURPOSE: Total kidney volume (TKV) is an important measure in renal disease detection and monitoring. We developed a fully automated method to segment the kidneys from T2 -weighted MRI to calculate TKV of healthy control (HC) and chronic kidney disease (CKD) patients. METHODS: This automated method uses machine learning, specifically a 2D convolutional neural network (CNN), to accurately segment the left and right kidneys from T2 -weighted MRI data. The data set consisted of 30 HC subjects and 30 CKD patients. The model was trained on 50 manually defined HC and CKD kidney segmentations. The model was subsequently evaluated on 50 test data sets, comprising data from 5 HCs and 5 CKD patients each scanned 5 times in a scan session to enable comparison of the precision of the CNN and manual segmentation of kidneys. RESULTS: The unseen test data processed by the 2D CNN had a mean Dice score of 0.93 ± 0.01. The difference between manual and automatically computed TKV was 1.2 ± 16.2 mL with a mean surface distance of 0.65 ± 0.21 mm. The variance in TKV measurements from repeat acquisitions on the same subject was significantly lower using the automated method compared to manual segmentation of the kidneys. CONCLUSION: The 2D CNN method provides fully automated segmentation of the left and right kidney and calculation of TKV in <10 s on a standard office computer, allowing high data throughput and is a freely available executable.

Quantitative assessment of renal structural and functional changes in chronic kidney disease using multi-parametric magnetic resonance imaging.

BACKGROUND: Multi-parametric magnetic resonance imaging (MRI) provides the potential for a more comprehensive non-invasive assessment of organ structure and function than individual MRI measures, but has not previously been comprehensively evaluated in chronic kidney disease (CKD). METHODS: We performed multi-parametric renal MRI in persons with CKD (n = 22, 61 ± 24 years) who had a renal biopsy and measured glomerular filtration rate (mGFR), and matched healthy volunteers (HV) (n = 22, 61 ± 25 years). Longitudinal relaxation time (T1), diffusion-weighted imaging, renal blood flow (phase contrast MRI), cortical perfusion (arterial spin labelling) and blood-oxygen-level-dependent relaxation rate (R2*) were evaluated. RESULTS: MRI evidenced excellent reproducibility in CKD (coefficient of variation <10%). Significant differences between CKD and HVs included cortical and corticomedullary difference (CMD) in T1, cortical and medullary apparent diffusion coefficient (ADC), renal artery blood flow and cortical perfusion. MRI measures correlated with kidney function in a combined CKD and HV analysis: estimated GFR correlated with cortical T1 (r = -0.68), T1 CMD (r = -0.62), cortical (r = 0.54) and medullary ADC (r = 0.49), renal artery flow (r = 0.78) and cortical perfusion (r = 0.81); log urine protein to creatinine ratio (UPCR) correlated with cortical T1 (r = 0.61), T1 CMD (r = 0.61), cortical (r = -0.45) and medullary ADC (r = -0.49), renal artery flow (r = -0.72) and cortical perfusion (r = -0.58). MRI measures (cortical T1 and ADC, T1 and ADC CMD, cortical perfusion) differed between low/high interstitial fibrosis groups at 30-40% fibrosis threshold. CONCLUSION: Comprehensive multi-parametric MRI is reproducible and correlates well with available measures of renal function and pathology. Larger longitudinal studies are warranted to evaluate its potential to stratify prognosis and response to therapy in CKD.

Multiparametric Renal Magnetic Resonance Imaging for Prediction and Annual Monitoring of the Progression of Chronic Kidney Disease over Two Years.

BACKGROUND: Multiparametric renal Magnetic Resonance Imaging (MRI) provides a non-invasive method to assess kidney structure and function, but longitudinal studies are limited. METHODS: A total of 22 patients with CKD category G3-4 (estimated glomerular filtration rate (eGFR) 15-59 mL/min/1.73 m2) were recruited. Annual 3T multiparametric renal MRI scans were performed, comprising total kidney volume (TKV), longitudinal relaxation time (T1), apparent diffusion coefficient (ADC), Arterial Spin Labelling, and Blood Oxygen Level Dependent relaxation time (T2*), with 15 patients completing a Year 2 scan. CKD progression over 2 years was defined as eGFR_slope ≥ -5 mL/min/1.73 m2/year. RESULTS: At baseline, T1 was higher (cortex p = 0.05, medulla p = 0.03) and cortex perfusion lower (p = 0.015) in participants with subsequent progression versus stable eGFR. A significant decrease in TKV and ADC and an increase in cortex T1 occurred in progressors at Year 1 and Year 2, with a significant decrease in perfusion in progressors only at Year 2. The only decline in the stable group was a reduction in TKV. There was no significant change in cortex or medulla T2* at Year 1 or Year 2 for progressors or stable participants. CONCLUSION: Lower renal cortex perfusion and higher T1 in the cortex and medulla may predict CKD progression, while renal cortex T1, TKV, and ADC may be useful to monitor progression. This study provides pilot data for future large-scale studies.