A new method for analyzing histograms of brain magnetization transfer ratios: comparison with existing techniques.

BACKGROUND AND PURPOSE Previously reported quantitative parameters for the magnetization transfer ratio (MTR) do not give identical results, which can limit their ability to differentiate normal from diseased tissue and render them vulnerable to variations among MR systems. Our purpose was to systematically study different MTR metrics; propose a new MTR histogram parameter, AMTR(2/3); and compare AMTR(2/3) with existing parameters in a study of multiple sclerosis (MS). METHODS Seven conventional MTR parameters were proposed: global and mean MTR; peak height and position of the histogram; and percentiles MTR25, MTR50, and MTR75. Additionally, we investigated a parameter, AMTR(2/3), to indicate the normalized pixel count (area under the histogram curve) inside the band size of two-thirds MTR histogram peak height. All parameters were measured in 10 patients with relapsing-remitting MS (group A), 10 healthy control subjects from the same imaging center as that of patients (group B), and four healthy control subjects from an outside institution (group C). Comparison of findings was performed between groups A and B to assess the discriminating ability of MTR parameters and groups B and C to evaluate intersystem variations. RESULTS All MTR parameters differed between groups A and B, but the difference was significant for only global MTR, mean MTR, MTR25, and AMTR(2/3). With the exception of AMTR(2/3), all parameters differed significantly between the two control groups. CONCLUSION AMTR(2/3) is less sensitive to MR imaging system variations than are other MTR parameters and was most effective in differentiating patients with MS from healthy control subjects. This finding supports the use of AMTR(2/3) in multicenter MT MR imaging studies of MS.