Minimization of errors in biexponential T2 measurements of the prostate.

Abstract

To determine the echo times that provide the greatest precision in measurements of prostate T2s. T2 relaxation time measurements in the prostate are complicated by the structure of prostate tissue, which consists of fluid-filled glands surrounded by epithelial and stromal cells. Since the glands are large relative to diffusion distances, there is little water exchange between the two compartments and T2s are biexponential. Because the relative size and characteristics of the two compartments change in prostate tumors, accurate measurement of the characteristics of each may provide useful information on tumor grade. T2s were measured in a group of 25 men with biopsy-proven prostate cancer. Subjects were scanned at 3T with a 16-echo turbo-spin echo T2-mapping sequence. Normal prostate T2s were measured in areas showing no disease. Optimum echo times for measurement of normal prostate T2s were found by calculating the covariance matrix, which provides estimates of parameter variance. Echo times that minimize T2 variance were then found by searching over grids of different echo times. Optima for four to eight echo acquisitions were found. Optima were tested by Monte Carlo simulation. Fast and slow T2s were 60 msec and 360 msec, respectively. The fast signal fraction was 0.6. Optimum echo times were between 0 and 780 msec, depending on the number of echoes acquired. Use of optimum echo times can substantially improve the precision of biexponential T2 measurements. This optimization is anticipated to improve prostate cancer characterization using T2 measurements.