Direct measurement of the signal intensity of diffusion-weighted magnetic resonance imaging for preoperative grading and treatment guidance for brain gliomas

Abstract

Background: Magnetic resonance diffusion-weighted imaging (DWI) has been widely used clinically in imaging diagnosis of intracranial disorders. The purpose of current study was to present a quantitative method of direct measuring the DWI signal intensity of brain gliomas on the monitors of hospital picture archiving and communicating system (PACS) for grading gliomas. Methods: This study recruited 135 patients with treatment-na?ve brain gliomas. Direct measurement of the signal intensity of selected tumoral regions of interest (ROIs) by DWI on the monitors of the hospital PACS was performed for all patients. From the measurements, we obtained three values, defined as DWIT (tumor), DWIN (the homologous normal-appearing area of the tumor ROI in the contralateral hemisphere), and DWIWM (normal-appearing white matter) in the contralateral frontal lobe. Two ratios, DWIT/WM and DWIT/N, were obtained for each tumoral ROI. The same method was used for apparent diffusion coefficient (ADC) ratios of the tumoral ROI. Fractional polynomial regression and the Mann-Whitney U test were applied to determine the correlation between tumor grading, MIB-1 labeling index, and DWI and ADC ratios. Logistic regression models and receiver operating characteristic curve analysis were used to establish diagnostic models. Measurements of intraobserver and interobserver agreement were also made at 1-month interval. Results: The DWI ratios correlated positively with tumor grade and MIB-1 value (p < 0.01). Cut-off ratios of 1.62 for DWIT/WM and 1.47 for DWIT/N generated the optimal combination of sensitivity (0.82, 0.80), specificity (0.79, 0.86), and sound discriminating power, with an area under the curve of 0.87 and 0.84, respectively, to differentiate low-grade from high-grade gliomas. ADC ratios showed relatively worse sensitivity, specificity, and discriminating power than DWI ratios. Almost all intraobserver and interobserver measurements were within 95% agreement. Conclusion: The proposed method - direct measuring of tumor signal intensity of DWI on PACS monitors - is feasible for grading gliomas in clinical neuro-oncology imaging services and has a high level of reliability and reproducibility. ? 2012 Elsevier Inc.