Reproducibility of Quantitative Histogram and Texture parameters from Intravoxel Incoherent Motion Diffusion-weighted MR Imaging in FN13762 Rat Breast Cancer Model

Abstract

Purpose: To determine the reproducibility of quantitative histogram and texture parameters from intravoxel incoherent motion (IVIM) diffusion-weighted magnetic resonance (MR) imaging in a FN13762 breast cancer in rats. Materials and Methods: This study was approved by the institutional animal care and use committee. FN13762 murine mammary carcinoma cell (5×106) were inoculated into the right upper thigh of Fisher 344 rats (n=11). At nine days after tumor implantation, IVIM diffusion-weighted MR imaging was performed twice with nine b values by using a 3-T MR imaging system. The entire tumors were manually segmented and their computerized histogram and texture parameters were automatically extracted from the apparent diffusion coefficient (ADC) and IVIM-derived parametric maps including true diffusion coefficient (Dt), pseudo-diffusion coefficient (Dp) and perfusion fraction (Pf) map using in-house software program. Intraobserver and interscan measurement reproducibility were evaluated using paired t-test, intraclass correlation coefficient (ICC) and Bland-Altman method. Any parameter that satisfied the lower limit of the 95% confidence interval (CI) for ICC ≥0.75 and absence of a systemic bias on both intraobserver and interscan comparison was designated as a reproducible parameter. Results: Volume measurements showed good intraobserver and interscan agreement (lower limit of the 95% CI for ICC, 0.97 and 0.95, respectively) without any systemic bias. The mean value, entropy, 5th, 10th, 25th and 90th percentile values from ADC map and the mean value, entropy, gray-level co-occurrence matrix (GLCM) contrast, 5th, 10th, 25th and 50th percentile values from Dt map showed good intraobserver and interscan agreement (lower limit of the 95% CI for ICC, ≥0.75) without any systemic bias. All parameters from Dp and Pf map showed relatively poor intraobserver and interscan agreement (lower limit of the 95% CI for ICC, <0.75). Conclusion: In conclusion, histogram and texture analysis on ADC and Dt map were more favorable than Dp and Pf map in terms of measurement reproducibility. Among histogram and texture parameters, the mean, entropy, 5th, 10th, 25th and 90th percentile value from ADC map and the mean, entropy, GLCM contrast, 5th, 10th, 25th and 50th percentile value from Dt map were reproducible.