Preoperative Prediction of Solitary Fibrous Tumor/Hemangiopericytoma and Angiomatous Meningioma Using Magnetic Resonance Imaging Texture Analysis.

Abstract

Solitary fibrous tumor (SFT)/hemangiopericytoma (HPC) is radiologically difficult to distinguish from meningioma, especially angiomatous meningioma. This study aimed to detect texture parameters to distinguish SFT/HPC from angiomatous meningioma using magnetic resonance imaging texture analysis with commercially available software. We retrospectively investigated textural parameters in 43 newly diagnosed SFTs/HPCs, angiomatous meningiomas, and other World Health Organization (WHO) grade I meningiomas treated at Keio University Hospital. For T1 contrast-enhanced, T2, and apparent diffusion coefficient (ADC) images, texture analyses were performed. Regions of interest were drawn manually with reference to the greater signal on contrast-enhanced T1-weighted images. ADC values and texture parameters, including kurtosis, skewness, and entropy, were evaluated and compared between these 3 groups. The mean ADC value was significantly high in angiomatous meningioma, compared with SFT/HPC and other WHO grade I meningioma. ADC entropy was highest in SFT/HPC, followed by angiomatous and other WHO grade I meningioma. T2 skewness was significantly high in SFT/HPC, compared with angiomatous and other WHO grade I meningioma. T1 contrast-enhanced skewness was significantly low in angiomatous meningioma, compared with other WHO grade I meningioma. Mean ADC value distinguished SFT/HPC from angiomatous meningioma with a positive predictive value (PPV) of 62.5% and specificity of 62.5%. ADC entropy distinguished SFT/HPC from angiomatous meningioma with a PPV of 100% and specificity of 100%. ADC skewness distinguished SFT/HPC from angiomatous meningioma with a PPV of 66.7% and specificity of 71.4%. This study demonstrated that magnetic resonance imaging texture analysis was useful for distinguishing SFT/HPC from meningioma, especially angiomatous meningioma.