Temperature measurement of intracranial cerebrospinal fluid using second-order motion compensation diffusion tensor imaging

To reduce the determination errors of CSF pulsation in diffusion-weighted image (DWI) thermometry, we investigated whether applying second-order motion compensation diffusion tensor imaging (2nd-MC DTI) and fractional anisotropy (FA) processing improves the measurement of intracranial cerebrospinal fluid (CSF) temperature. In a phantom study, we investigated the relationship between temperature and FA in artificial CSF (ACSF) to determine the threshold for FA processing. The calculated temperatures of ACSF were compared with those of water. In a human study, 18 healthy volunteers were scanned using conventional DTI (c-DTI) and 2nd-MC DTI on a 3.0 T magnetic resonance imaging (MRI) system. A temperature map was created using diffusion coefficients from each DWI with/without FA processing. The temperatures of intracranial CSF were compared between each DTI image using Welch's analysis of variance and Games-Howell's multiple comparisons. In the phantom study, FA did not exceed 0.1 at any temperature. Consequently, pixels exceeding the threshold of 0.1 were removed from the temperature map. Intracranial CSF temperatures significantly differed between the four methods (p < 0.0001). The lowest temperature was 2nd-MC DTI with FA processing (mean, 35.62 degrees C), followed in order by c-DTI with FA processing (mean, 36.16 degrees C), 2nd-MC DTI (mean, 37.08 degrees C), and c-DTI (mean, 39.08 degrees C; p < 0.01 for each). Because the calculated temperature of ACSF was estimated to be lower than that of water, the temperature of 2nd-DTI with FA processing was considered reasonable. The method of 2nd-MC DTI with FA processing enabled determining intracranial CSF temperature with a reduction in CSF pulsation.