Dataset Information

Measurement of Glomerular Filtration Rate using Quantitative SPECT/CT and Deep-learning-based Kidney Segmentation.

ABSTRACT: Quantitative SPECT/CT is potentially useful for more accurate and reliable measurement of glomerular filtration rate (GFR) than conventional planar scintigraphy. However, manual drawing of a volume of interest (VOI) on renal parenchyma in CT images is a labor-intensive and time-consuming task. The aim of this study is to develop a fully automated GFR quantification method based on a deep learning approach to the 3D segmentation of kidney parenchyma in CT. We automatically segmented the kidneys in CT images using the proposed method with remarkably high Dice similarity coefficient relative to the manual segmentation (mean?=?0.89). The GFR values derived using manual and automatic segmentation methods were strongly correlated (R2?=?0.96). The absolute difference between the individual GFR values using manual and automatic methods was only 2.90%. Moreover, the two segmentation methods had comparable performance in the urolithiasis patients and kidney donors. Furthermore, both segmentation modalities showed significantly decreased individual GFR in symptomatic kidneys compared with the normal or asymptomatic kidney groups. The proposed approach enables fast and accurate GFR measurement.

SUBMITTER: Park J

PROVIDER: S-EPMC6414660 | biostudies-literature | 2019 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Measurement of Glomerular Filtration Rate using Quantitative SPECT/CT and Deep-learning-based Kidney Segmentation.

Park Junyoung J Bae Sungwoo S Seo Seongho S Park Sohyun S Bang Ji-In JI Han Jeong Hee JH Lee Won Woo WW Lee Jae Sung JS

Scientific reports 20190312 1

Quantitative SPECT/CT is potentially useful for more accurate and reliable measurement of glomerular filtration rate (GFR) than conventional planar scintigraphy. However, manual drawing of a volume of interest (VOI) on renal parenchyma in CT images is a labor-intensive and time-consuming task. The aim of this study is to develop a fully automated GFR quantification method based on a deep learning approach to the 3D segmentation of kidney parenchyma in CT. We automatically segmented the kidneys i ...[more]

PMID: 30862873

Similar Datasets

Project description:PurposeTo assess the accuracy of glomerular filtration rate (GFR) measurements obtained with low-contrast agent dose dynamic contrast material-enhanced magnetic resonance (MR) renography in patients with liver cirrhosis who underwent routine liver MR imaging, with urinary clearance of technetium 99m ((99m)Tc) pentetic acid (DTPA) as the reference standard.Materials and methodsThis HIPAA-compliant study was institutional review board approved. Written informed patient consent was obtained. Twenty patients with cirrhosis (14 men, six women; age range, 41-70 years; mean age, 54.6 years) who were scheduled for routine 1.5-T liver MR examinations to screen for hepatocellular carcinoma during a 6-month period were prospectively included. Five-minute MR renography with a 3-mL dose of gadoteridol was performed instead of a routine test-dose timing examination. The GFR was estimated at MR imaging with use of two kinetic models. In one model, only the signal intensities in the aorta and kidney parenchyma were considered, and in the other, renal cortical and medullary signal intensities were treated separately. The GFR was also calculated by using serum creatinine levels according to the Cockcroft-Gault and modification of diet in renal disease (MDRD) formulas. All patients underwent a (99m)Tc-DTPA urinary clearance examination on the same day to obtain a reference GFR measurement. The accuracies of all MR- and creatinine-based GFR estimations were compared by using Wilcoxon signed rank tests.ResultsThe mean reference GFR, based on (99m)Tc-DTPA clearance, was 74.9 mL/min/1.73 m(2) ± 27.7 (standard deviation) (range, 10.3-120.7 mL/min/1.73 m(2)). With both kinetic models, 95% of MR-based GFRs were within 30% of the reference values, whereas only 40% and 60% of Cockcroft-Gault- and MDRD-based GFRs, respectively, were within this range. MR-based GFR estimates were significantly more accurate than creatinine level-based estimates (P < .001).ConclusionGFR assessment with MR imaging, which outperformed the Cockcroft-Gault and MDRD formulas, adds less than 10 minutes of table time to a clinically indicated liver MR examination without ionizing radiation.Supplemental materialhttp://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.11101338/-/DC1.

Project description:BackgroundEndogenous molecules that provide an unbiased and a precise evaluation of kidney function are still necessary. We explored the potential of clearance of d-serine, a rare enantiomer of serine and a biomarker of kidney function, as a measure of glomerular filtration rate (GFR).MethodsThis was a cross-sectional observational study of 200 living kidney transplant donors and recipients enrolled between July 2019 and December 2020 in a single Japanese center, for whom GFR was measured by clearance of inulin (C-in). Clearance of d-serine (C-dSer) was calculated based on blood and urine levels of d-serine, as measured by two-dimensional high-performance liquid chromatography. Analytical performance was assessed by calculating biases. Utilizing data from 129 participants, we developed equations for C-in based on C-dSer and C-cre using a linear regression model, and the performance was validated in 68 participants.FindingsThe means of C-in and C-dSer were 66.7 and 55.7 mL/min/1.73 m2 of body surface area, respectively, in the entire cohort. C-dSer underestimated C-in with a proportional bias of 22.0% (95% confidence interval, 14.2-29.8%) and a constant bias of -1.24 (-5.78-3.31), whereas the proportional bias was minor to that of C-cre (34.6% [31.1-38.2%] and 2.47 (-1.18-6.13) for proportional and constant bias, respectively). Combination of C-dSer and C-cre measured C-in with an equation of 0.391 × C-dSer + 0.418 × C-cre + 3.852, which reduced the proportional bias (6.5% [-0.2-13.1%] and -4.30 [-8.87-0.28] for proportional and constant bias, respectively). In the validation dataset, this equation performed well with median absolute residual of 3.5 [2.3-4.8], and high ratio of agreement (ratios of 30% and 15% different from C-in [P30 and P15] of 98.5 [91.4-100] and 89.7 [80.0-95.2], respectively).InterpretationThe smaller proportional bias compared to that of C-cre is an advantage of C-dSer as a measure of C-in. Combinational measurement of d-serine and creatinine, two endogenous molecules, has the potential to serve as a measure of GFR with precision and minor biases and can support important clinical decisions.FundingJapan Society for the Promotion of Science (JSPS, grant number 17H04188), Japan Agency of Medical Research and Development (AMED, JP20gm5010001), Osaka Kidney Bank (OKF19-0010), Shiseido Co., Ltd and KAGAMI Inc.

Project description:Background and objectivesThe tracing of creatinine (Cr) reference materials to isotope dilution mass spectrometry-assigned values was implemented worldwide to reduce interlaboratory variability and improve assay accuracy. The aims of this study were to examine the current extent of interlaboratory variability and its effect on eGFR.Design, setting, participants, & measurementsLeftover plasma from 2-3 consecutive days was obtained from 53 intensive care unit patients with a range of kidney functions. Individual patient samples were pooled and split and sent to 12 different laboratories for Cr measurement. For each patient, the mean Cr and Chronic Kidney Disease Epidemiology Collaboration eGFR (eGFR-EPI), assuming a 65-year-old nonblack woman, were determined. Interlaboratory variability was assessed by the range and SD of Cr and eGFR-EPI. This was repeated after stratifying by assay type and by the median Cr of 1.36 mg/dl. For patients whose eGFR-EPI range included 60 ml/min per 1.73 m2, the percentage of laboratories with eGFR-EPI<60 ml/min per 1.73 m2 was determined.ResultsThe mean±SD of the Cr and eGFR-EPI ranges were 0.20±0.09 mg/dl and 14±9 ml/min per 1.73 m2 for Cr<1.36 mg/dl. Jaffe Cr results were an average 0.1 mg/dl (Cr?1.36 mg/dl) and 0.05 mg/dl (Cr<1.36 mg/dl) higher than enzymatic results (P<0.001 for both). Ten patients had an eGFR-EPI range that included 60 ml/min per 1.73 m2. Their median eGFR-EPI range was 15 ml/min per 1.73 m2. There was significant discordance in the diagnosis of CKD (eGFR-EPI<60 ml/min per 1.73 m2), with laboratories using Jaffe Cr methods making the diagnosis more frequently than those using enzymatic Cr methods (60% versus 39%).ConclusionsSignificant interlaboratory variability in Cr measurement still exists. Jaffe assays yield higher Cr values than enzymatic assays, leading to lower eGFR-EPIs and more frequent CKD diagnoses. Further improvements in assay performance are required to standardize patient CKD diagnosis and to facilitate longitudinal Cr monitoring across laboratories.

Dataset Information

Measurement of Glomerular Filtration Rate using Quantitative SPECT/CT and Deep-learning-based Kidney Segmentation.

Publications

Measurement of Glomerular Filtration Rate using Quantitative SPECT/CT and Deep-learning-based Kidney Segmentation.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets