Twenty-one in vivo studies of bone marrow of the lumbar spine were performed with a 0.6-T commercial MR imager and proton chemical shift imaging techniques. Six healthy volunteers served as controls. Multiple measurements in the volunteers demonstrated reproducibility within errors of 5% for fat fraction and 6% for T1 of water. Ten patients who had histologically proved leukemia or aplastic anemia were then examined. The data show that changes in fat fraction represent the underlying reason for many of the changes observed in conventional spin-echo (SE) images of these disorders. Although both conventional and chemical shift images showed differences among the pathologic groups and healthy volunteers, fat fraction determined with chemical shift imaging was the single best discriminator among them. A two-point estimate of fat fraction was also evaluated. This rapid imaging protocol performed almost as well as the complete quantitative analysis in discriminating between pathologic and healthy tissue and showed improved discrimination compared with conventional SE techniques.

No abstract

Magnetic

To determine the accuracy of left ventricular ejection fractions (EFs) calculated from magnetic resonance (MR) images, 22 patients who underwent coronary angiography and left ventriculography were studied within 1-3 days by MR imaging. ECG-gated spin-echo 30-msec echo-delay images were obtained in end systole and end diastole in a plane through the long axis of the left ventricle perpendicular to the septum at a level through the aortic valve and apex. The area-length method was then used to calculate the EF from left ventriculograms and MR images. Angiographic EF correlated linearly with MR determinations. The mean differences in EF calculation between two observers were 5.0 +/- 6.9 for ventriculography and 2.7 +/- 7.8 for MR imaging. MR imaging, using image planes oriented to the left ventricular axes, can allow accurate EF calculation. This has the potential to be clinically useful in noninvasive evaluation of cardiac function.

No abstract