Detection of Kidney Impairment Using Correlation of CT Number and ADC

Abstract

<h3>Purpose/Objective(s)</h3> It has been reported that a correlation exists between the mean apparent diffusion coefficient (ADC) measured from MRI and the mean CT number (CTN) in kidney. This correlation can be interpreted as both ADC and CTN reflect extracellular volume (ECV) ratio. This study aims to investigate the correlation of ADC and estimated glomerular filtration ratio (eGFR) and the use of ADC-CTN correlation to detect kidney impairment. <h3>Materials/Methods</h3> Data acquired for radiation therapy (RT) planning of twenty-two patients with abdominal/pelvic cancers were analyzed retrospectively in two groups. Group I included sixteen patients with no history of severe kidney disease and Group II had six patients having apparent issues including kidney stones, chronic disease, or tumor on kidney. CT images were acquired during CT simulation using a scanner at 120 kVp and the ADC maps were derived from the diffusion weighted imaging (DWI) acquired immediately before the first fraction of MRI guided RT using a 1.5T MR-Linac approximately 3 weeks following CT simulation. Left and right kidneys excluding renal pelvis were contoured on both the ADC and CT images. The mean values of the ADC and CTN from the voxels within the contours were analyzed using linear regression. The eGFR was calculated based on serum creatinine level measured on the same day as DWI. The correlation between eGFR and mean ADC value of left and right kidneys was evaluated with Pearson test. The linear regression fit was used to correct the ADC values to eliminate the filtration effect to show the ECV effect only. <h3>Results</h3> The CTN and ADC over the Group I patients had an average 29.3 HU and 2.00 µm/s² ranging from 23.0 HU to 38.2 HU and 1.79 µm/s² to 2.16 µm/s², respectively. The eGFR of all patients ranged from 40 to 129 mL/min/1.73 sqm, and had a moderate correlation to ADC (coefficient=0.46, p=0.04). Using the linear fit of ADC-eGFR data, the ADC values were corrected by extrapolating their corresponding eGFR to 0. This correction did not alter the strong correlation between the ADC and CTN (Pearson coefficient = -0.83 with and -0.84 without correction, p<0.0001) but dropped the line of linear fit by 0.2 µm/s². There was no significant difference in eGFR, CTN, or ADC between Group I and II. However, almost all patients in Group II (5 out of 6) had one kidney data residing within the +/- 0.15 µm/s² region of fit line while the other one fell out of the region as an outlier, identified as the actual problematic kidney in the medical record (4 of 5). In two cases kidney stones were not found when the images were taken but in 6-12 months posteriorly. <h3>Conclusion</h3> Our study showed that eGFR affected ADC in a minor way and would not change the strong negative correlation between ADC and CTN. Impaired kidneys tend to fall out of the ADC-CTN correlation line as outliers, which were not identified using eGFR, CTN, or ADC alone. These results indicate that ADC-CTN correlation is a new sensitive method to detect kidney impairment and has a potential application in monitoring radiation toxicity.