Purpose
To assess the added value of contrast‐enhanced (CE) MR sequences (static CE‐MR sequences, dynamic CE‐MR sequences) to noncontrast enhanced MR sequences (non‐CE‐MR sequences) including T1, fluid‐sensitive, and diffusion‐weighted imaging (DWI) with apparent diffusion coefficient (ADC) mapping for characterizing “indeterminate” soft tissue masses (STMs) as benign or malignant.
Materials and Methods
Thirty‐nine patients with indeterminate STMs (27 benign, 12 malignant) underwent 3 Tesla MRI with conventional non‐CE‐MR sequences (T1‐weighted, fluid‐sensitive), DWI (b‐values 50, 400, 800, ADC mapping), dynamic CE‐MR sequences (7‐s time resolution), and static CE‐MR sequences. Two readers independently reviewed imaging in four sessions (conventional non‐CE‐MR sequences alone, conventional+DWI/ADC, conventional+DWI/ADC+static CE‐MR sequences, conventional+DWI/ADC+static CE‐MR sequences dynamic CE‐MR sequences). Readers recorded the potential of malignancy at each session; reader diagnostic performance (receiver operating characteristics analysis) and inter‐observer variability (weighted kappa [k]) were determined.
Results
Diagnostic performance for distinguishing benign and malignant STMs was highest with the addition of dynamic CE‐MR sequences (reader 1, area under the curve [AUC] 0.91; reader 2, AUC 0.88). The diagnostic performance of static CE‐MR sequences (reader 1, AUC 0.86; reader 2, AUC 0.84) was not superior to non‐CE‐MR sequences with DWI (reader 1, AUC 0.88; reader 2, AUC 0.8). Interobserver agreement was: k = 0.82 (static CE‐MRI), k = 0.79 (dynamic CE‐MRI), k = 0.53 (non‐CE‐MR sequences without DWI), and k = 0.63 (with DWI).
Conclusion
Non‐CE‐MR sequences offer similar diagnostic performance to imaging with the addition of static CE‐MR sequences, but their interobserver reliability is lower. The addition of dynamic CE‐MR sequences offers the higher diagnostic performance for distinguishing benign and malignant indeterminate STMs.
Level of Evidence: 3
J. Magn. Reson. Imaging 2017;45:390–400.