Dg. Disler et al., IN-PHASE AND OUT-OF-PHASE MR-IMAGING OF BONE-MARROW - PREDICTION OF NEOPLASIA BASED ON THE DETECTION OF COEXISTENT FAT AND WATER, American journal of roentgenology, 169(5), 1997, pp. 1439-1447
OBJECTIVE. The purpose of this study was to determine if gradient-echo
MR imaging with TEs selected with fat and water in phase and out of p
hase can help predict the likelihood of neoplastic or nonneoplastic le
sions in bone marrow. SUBJECTS AND METHODS. Thirty consecutive patient
s with 31 suspected bone marrow lesions underwent MR imaging, includin
g two spoiled gradient-echo sequences identical in all parameters exce
pt TE, which was chosen such that fat and water were either in phase o
r out of phase. Relative ratios of the abnormal bone marrow signal int
ensity and a control site on the in-phase and out-of-phase images were
expressed. The images were also assessed independently by mio reviewe
rs who were unaware of the patients' identities and clinical histories
. Reviewers assessed decreased marrow signal intensity relative to con
trol sites on the out-of-phase and in-phase images. Pathologic confirm
ation was obtained in 16 patients (17 lesions); the remainder of patie
nts had either established diagnoses or determination of benignity bas
ed on stability of findings at 1 year. Relative ratios were compared w
ith the Student's t test and receiver operating characteristic (ROC) c
urve analysis, and the reviewers' scores were evaluated with ROC curve
analysis. RESULTS. The relative signal-intensity ratios were 1.03 +/-
0.13 for the neoplastic group and 0.62 +/- 0.13 for the nonneoplastic
group (p < .0001). ROC curve analysis of the signal-intensity ratios
showed a z-score of .99. A ratio cutoff value of 0.81 resulted in a 95
% sensitivity and a 95% specificity for detection of neoplasm. Both re
viewers achieved 100% sensitivity and 94-100% specificity for detectio
n of neoplasms. CONCLUSION. In-phase and out-of-phase gradient-echo MR
imaging of bone marrow signal-intensity abnormalities can help predic
t the likelihood of neoplastic or nonneoplastic lesions.