Application of multi-model diffusion weighted imaging in preoperative grading diagnosis of glioma

Objective To investigate the applicability and correlation of diffusion weighted imaging (DWI) based on monoexponential, biexponential and stretched-exponential models for preoperative grading of glioma.Methods A retrospective analysis was performed based on imaging data from 31 brain glioma patients. The corresponding parameter values of monoexponential, biexponential and stretched-exponential models were calculated. Then, the values of standard apparent diffusion coefficient (ADC), fast ADC, slow ADC, water diffusion heterogeneityindex (α) and distributed diffusion coefficient (DDC) were recorded and analyzed.Results The values of standard ADC, slow ADC, α, DDC decreased with the increase in the grade of brain glioma, while the fast ADC increased with higher glioma grades. There was a significant positive correlation between standard ADC and slow ADC, α, DDC respectively. Additionally, a positive correlation was observed between slow ADC and DDC, slow ADC and α. The standard ADC, slow ADC, α, and DDC values were negatively correlated with brain glioma grades, whereas fast ADC showed a positive correlation with glioma grades. The ROC analysis revealed that for diagnosing low-grade brain gliomas, the area under the curve (AUC) for standard ADC, DDC and α was at 0.878, 0.826, and 0.809, respectively,which were relatively large. The AUC for fast ADC in diagnosing high-grade brain gliomas was 0.819.Conclusions DWI based on monoexponential, biexponential and stretched-exponential models is an effective tool for the preoperative grading of brain gliomas.