Effectiveness of MR imaging in characterizing small hepatic lesions: routine versus expert interpretation.

OBJECTIVE

The aim of our study was to compare the effectiveness of MR imaging characterization of small (<or=2 cm) hepatic lesions made in a routine clinical setting with the effectiveness of such characterization made under standardized conditions by radiologists who are expert interpreters of MR imaging.

MATERIALS AND

METHODS

Forty-eight patients with 69 small (<or=2 cm) hepatic lesions considered indeterminate on a prior routine CT scan were included in the study. The diagnosis for all lesions had been verified by histology (n = 10), surgery and intraoperative sonography (n = 5), imaging follow-up (n = 35), or clinical follow-up (n = 19). Using the initial radiology reports, the diagnoses based on MR imaging were rated on a 5-point confidence scale. In addition, two radiologists experienced in MR imaging who were unaware of the initial interpretations of the images or the clinical histories of the patients independently analyzed the MR imaging studies and characterized the lesions using the same 5-point scale. The observer performance for the initial MR imaging interpretations and the expert interpretations were measured using receiver operating characteristic analysis. Interobserver agreement was determined with weighted kappa statistics.

RESULTS

Fifty-eight lesions were benign (six cysts, 22 hemangiomas, four regenerating nodules, two steatohepatitic lesions, one atypical blood vessel, three focal fat and five focal fat-sparing lesions, 13 flow-related pseudolesions, one diaphragmatic insertion, and one unspecified lesion), and 11 lesions were malignant (nine metastases and two hepatocellular carcinoma). The areas under the curve were 0.94 (initial reports), 0.88 (observer 1), and 0.84 (observer 2). Substantial agreement was found between the expert interpreters (kappa = 0.74), and moderate agreement, between the expert interpreters and initial interpreters (kappa = 0.44 each).

CONCLUSION

MR imaging is an effective method of characterizing small (<or=2 cm) hepatic lesions in routine clinical practice.