Differentiation between malignant and benign musculoskeletal tumors using diffusion kurtosis imaging.

Abstract

The purpose of this study was to evaluate differences in parameters of diffusion kurtosis imaging (DKI) and minimum apparent diffusion coefficient (ADCmin) between benign and malignant musculoskeletal tumors. In this prospective study, 43 patients were scanned using a DKI protocol on a 3-T MR scanner. Eligibility criteria were: non-fatty, non-cystic soft tissue or osteolytic tumors; >2cm; location in the retroperitoneum, pelvis, leg, or neck; and no prior treatment. They were clinically or histologically diagnosed as benign (n = 27) or malignant (n = 16). In the DKI protocol, diffusion-weighted imaging was performed using four b values (0-2000s/mm2) and 21 diffusion directions. Mean kurtosis (MK) values were calculated on the MR console. A recently developed software application enabling reliable calculation was used for DKI analysis. MK showed a strong correction with ADCmin (Spearman's rs = 0.95). Both MK and ADCmin values differed between benign and malignant tumors (p < 0.01). For benign and malignant tumors, the mean MK values (±SD) were 0.49 ± 0.17 and 1.14 ± 0.30, respectively, and ADCmin values were 1.54 ± 0.47 and 0.49 ± 0.17 × 10-3mm2/s, respectively. At cutoffs of MK = 0.81 and ADCmin = 0.77 × 10-3mm2/s, the specificity and sensitivity for diagnosis of malignant tumors were 96.3 and 93.8% for MK and 96.3 and 93.8% for ADCmin, respectively. The areas under the curve were 0.97 and 0.99 for MK and ADCmin, respectively (p = 0.31). MK and ADCmin showed high diagnostic accuracy and strong correlation, reflecting the accuracy of MK. However, no clear added value of DKI could be demonstrated in differentiating musculoskeletal tumors.