Radiologists and clinicians interested in differentiating malignant and benign vertebral bone marrow diseaseThe aim of this study is to evaluate the value of intravoxel incoherent motion (IVIM) for the differentiation between benign and malignant vertebral bone marrow lesions.This study was approved by local ethics committee. 13 patients (age: 31-79 years, M/F:7/6) with spinal diseases confirmed pathologically or clinically were recruited into the study. They were divided into three groups: tuberculous spondylitis(10 vertebras), benign edema (simple acute compression fracture ,6 vertebras ), and malignant tumor (10 vertebras). All patients underwent IVIM-MRI on a 1.5T scanner (Achieva 1.5T, Philips Healthcare, Best, The Netherlands) with an 15-channel SENSE spinal coil. The IVIM protocol was performed together with proton density-weighted spectral attenuated inversion recovery (PDW-SPAIR) with 8 b values of b =10, 20, 40, 80, 100, 200, 400, 600 s/mm², respectively. IVIM parameters diffusion coefficients (D) , perfusion fractions (f) and pseudo-diffusion coefficient (D * ) were calculated using a bi-exponential model on a pixel by pixel basis. Conventional high-b-value DWI sequence (0/600 s/mm²) was also performed to obtain the apparent diffusion coefficient (ADC) using a mono-exponential model. The ROI measurements of IVIM parametric maps were obtained with Image J(National Institutes of Health, Bethesda, MD, USA). The average of D, f, D* and ADC among the three groups were compared statistically by using one-way analysis of variance (ANOVA). Receiver operating characteristic (ROC) was conducted to compare the diagnostic performance of IVIM-MRI and DWI-MRI for differentiating bone marrow benign from malignant lesions. P value <0.05 was considered significant.

The average of D, f, D* and ADC among the three groups are presented in Table 1. The results showed malignant tumor group had a statistically lower D and ADC values than that of tuberculous spondylitis and benign edema groups (P<0.01). But the malignant tumor had a statistically higher D* value than that of the other two groups (P<0.05). In addition, Benign edema had significantly higher f value than that of the other two groups (P<0.01). ROC curves showed the AUC of D, f, ADC for diagnosis of benign lesions and malignancy lesions were 0.981, 0.791, and 0.925, respectively. This results hinted that the value of D, D* and ADC can be helpful to differentiate vertebral benign lesions (tuberculous spondylitis or acute compression fracture) from vertebral malignancy lesions. However, they could not distinguish benign edema from tuberculous spondylitis. The f value may be helpful for the differentiation between benign edema and tuberculous spondylitis. This study demonstrates that IVIM can be used to evaluate both diffusion and perfusion characteristics of spinal benign or malignant diseases.

IVIM is potentially a promising and valuable non-invasive method in differentiating the spinal benign and malignant diseases.