Apparent diffusion coefficient (ADC) measurement in endometrial carcinoma: Effect of region of interest methods on ADC values

Abstract

To investigate the influence of different-shaped regions of interest (ROIs) on tumor apparent diffusion coefficient (ADC) measurements and interobserver variability in endometrial carcinoma. Sixty-nine patients (age range, 32-92 years; mean, 61 years) were evaluated in this retrospective study. Patients had undergone magnetic resonance (MR) examinations including diffusion-weighted imaging (DWI) using a 3.0-T MR system. Two readers measured tumor ADCs using four ROI methods: freehand ROI; square ROI; round ROI; and five small, round ROIs. Minimum and mean ADCs were obtained. The interclass correlation coefficient (ICC) was statistically analyzed to assess measurement reliability. Repeated-measures analysis of variance was used for comparisons of ADCs measured with each ROI method. ICCs were 0.93 (minimum ADC) / 0.93 (mean ADC) for freehand ROIs, 0.94/0.95 for square ROIs, 0.94/0.95 for round ROIs, and 0.95/0.96 for five small, round ROIs. All ROI methods indicated excellent correlations. Each minimum ADC was significantly different except between square ROI and round ROI (P < 0.001). Mean ADCs showed significant differences only between freehand ROI and the other ROI methods (P < 0.001). ROI shape has no marked influence on ICC in endometrial carcinoma. Compared with minimum ADCs, mean ADCs are suggested to provide more stable results regardless of the ROI method.