Comparison between different implementations of the 3D FLASH sequence for knee cartilage quantification

Abstract

To compare several sequence implementations of the 3D FLASH sequence in the context of quantitative cartilage imaging. Test-retest coronal fast low angle shot (FLASH) sequences with water excitation were acquired in knees of 12 healthy participants, using two 1.5 T scanners from the same manufacturer. On one of the scanners, the FLASH was additionally compared with a FLASH VIBE, 75% with 100% slice resolution, a non-selective with a conventional spatial pulse, and "asymmetric echo allowed" with "not allowed". Implementations of the FLASH showed systematic differences of up to 3.3%, but these were not statistically significant. Precision errors were similar between protocols, but tended to be smallest for the FLASH VIBE with 100% slice resolution (0.6-6.7%). In the medial tibia cartilage volume and thickness differed significantly (P < 0.01; 6.2 and 5.9%) between the two scanners. Using a validated FLASH sequence, one can reduce slice resolution to 75% and allow asymmetric echo without sacrificing precision, in order to reduce the total acquisition time. However, in longitudinal studies, the scanner and the specific sequence implementation should be kept constant between baseline and follow-up, in order to avoid systematic off-sets in the measurements.