Diagnostic accuracy of diffusion tensor imaging in differentiating malignant from benign compressed vertebrae

Abstract

Purpose: To study diagnostic accuracy of diffusion tensor imaging (DTI) in differentiating malignant from benign compressed vertebrae. Methods: This study was done on 43 patients with compressed vertebrae on conventional magnetic resonance study that underwent DTI. The mean diffusivity (MD) and fractional anisotropy (FA) of malignant (n = 24) and benign (n = 19) compressed vertebrae were calculated by two readers. Results: There was a significantly lower (P = 0.001) MD of both readers between malignant (0.74 ± 0.2 and 0.78 ± 0.2 × 10−3 mm2/s) and benign (1.67 + 0.3 and 1.63 ± 0.3 × 10−3 mm2/s) compressed vertebrae. The FA of malignant compressed vertebrae of both readers (0.55 ± 0.2 and 0.52 ± 0.1) was significantly higher (P = 0.001) than that of benign (0.26 ± 0.1 and 0.28 ± 0.1) compressed vertebrae. There was excellent inter-reader agreement between both readers using MD (K = 0.91) and FA (K = 0.86). The thresholds of MD and FA used for differentiating malignant from benign compressed vertebrae of both readers were 1.15 and 1.16 × 10−3 mm2/s and 0.37 and 0.34 with area under the curve (AUC) of 0.98, 0.96, 0.93, and 0.92 and diagnostic accuracy of 95.3%, 88.4%, 90.1%, and 86.0% respectively. Combined MD and FA revealed AUC of 0.99 and 0.97 and diagnostic accuracy of 95.3% and 93.0% by both readers respectively. Conclusion: DTI is a non-invasive technique providing accurate imaging parameters that can be used for differentiating malignant from benign compressed vertebrae.