Measurement of Fat Content in Vertebral Marrow Using a Modified Dixon Sequence to Differentiate Benign From Malignant Processes

Purpose: To determine whether fat-signal-fraction (FF) map using a modified Dixon sequence could help differentiate benign from malignant bone lesions. Materials and Methods: Spine magnetic resonance images (MRIs) of 120 consecutive patients were studied by using a 3T MRI with standard T1-weighted image (Ti WI) and modified-Dixon sequence for FF measurement. There were three groups: a control group (n 51) with normal vertebrae; a benign group (n = 40) with focal red marrow deposition, Schmorl's nodes, benign compression fracture, or Modic type 1 endplate degeneration; a malignant group (n = 29) with spinal malignancies. The following three parameters were measured on T1WI and FF map by two radiologists independently: T1 signal intensity (SI), FF and T1 SI of normal disc (SI). Then, Lesion-to-disc ratio (LDR = SI of the lesion/Sid) and FF ratio of lesion and normal marrow were calculated. The mean values of the parameters were compared among the groups and Receiver Operating Characteristic (ROC) curves were analyzed. Then a logistic regression was performed. Results: The FF (2.8%) and FF ratio (0.082) of malignancy were lower than benign lesions (P < 0.001). There was no difference in the LDR between malignancy and Schmorl's nodes (P= 0.795) or a benign compression fracture (P= 0.866). The areas under the ROC curves of FF and FF ratio were 93% and 87%, respectively, which were higher than those of the other parameters used for differentiation (P < 0.001). In logistic regression analyses, FF remained a significant variable that could be used to independently differentiate benign from malignant lesions, with an odds ratio of 1.9 (P < 0.001). Conclusion: The FF and FF ratio obtained from FF maps using modified-Dixon sequence could be used to distinguish between benign and malignant causes of focal bone marrow abnormalities when difficulty in the qualitative interpretation of conventional MR images arises.