T2 estimates in healthy and diseased brain tissue: a comparison using various MR pulse sequences

Fourteen patients and five healthy individuals underwent magnetic resonance (MR) imaging to determine an effective multiple spin echo pulse sequence for estimating T2. Lesions examined included infarction, glioma, multiple sclerosis, and acute hematoma. A pulse repetition time (TR) of 1,500 msec and echo delays (TEs) of 25, 50, 75, and 100 msec were used. Computed T2 images were derived from all four echoes, the first two echoes, and the first and fourth echoes. T2 values were obtained from specific brain locales using region-of-interest analysis. Use of either the first two echoes or the first and fourth in the T2 fit provided T2 estimates which closely correlated with that of the four-echo analysis. The noise level in T2 maps constructed from the 25- and 100-msec echoes was modestly (typically 10%) higher than that from four echoes; noise level from the 25- and 50-msec echoes was markedly higher, typically 60%. This behavior is remarkably consistent with that predicted from theory. All 19 subjects displayed consistent relative T2 values for specific brain structures; in 13, the absolute T2 values fell within a limited range. Despite the high sensitivity of T2 images, their specificity in the detection of most brain disease appears limited except in acute intracerebral hematoma, which exhibited a decreased T2 relaxation time using high-field-strength MR imaging.