Signal characteristics of focal liver lesions on double echo T2-weighted conventional spin echo MRI: observer performance versus quantitative measurements of T2 relaxation times.

PURPOSE The purpose of this work was to evaluate the ability of expert readers to differentiate benign from malignant liver lesions based on visual assessment of lesion signal intensity on double echo T2-weighted conventional spin echo (CSE) MR images and to compare reader performance with quantitative measurements of T2 relaxation times. METHOD Sixty-seven MR examinations demonstrating 85 liver lesions (37 hemangiomas, 32 malignancies, 15 cysts, and 1 focal nodular hyperplasia) on double echo T2-weighted CSE sequences (TR 3,600 ms/TE 50, 160 ms) were qualitatively reviewed by three independent readers. T2 relaxation times were calculated for each lesion. Receiver operating characteristic (ROC) analyses of expert readers were compared with calculated T2 relaxation times. RESULTS T2 values performed significantly better than subjective reader analysis for liver lesion characterization (area under ROC = 0.93 vs. 0.81, 0.78, and 0.75; p < 0.0001). With use of a T2 threshold of 125 ms, the sensitivity of T2 values for malignant lesions was 100%, specificity 71%, and accuracy 84%. By comparison, the sensitivity of the three readers for malignant lesions was 76-83%, with a specificity of 61-72% and an overall accuracy of 71-80%. CONCLUSION Despite expert reader analyses, subjective evaluations of liver lesion signal characteristics are prone to inaccuracy and lack certainty and consistency when intermediate TEs (50/160 ms) are used. Quantitative measurements of T2 relaxation times should be performed to accurately and confidently differentiate benign from malignant liver lesions. Use of a higher T2 threshold than previously recommended is required to avoid misclassification of malignancies.