Erschienen in:
01.08.2019 | Original Communication
Optimizing 3D FLAIR to detect MS lesions: pushing past factory settings for precise results
verfasst von:
Augustin Lecler, C. Bouzad, R. Deschamps, F. Maizeroi, J. C. Sadik, A. Gueguen, O. Gout, H. Picard, J. Savatovsky
Erschienen in:
Journal of Neurology
|
Ausgabe 11/2019
Einloggen, um Zugang zu erhalten
Abstract
Background
To assess the diagnostic value of three 3D FLAIR sequences with differing repetition-times (TR) at 3-Tesla when detecting multiple sclerosis (MS) lesions.
Methods
In this prospective study, approved by the institutional review board, 27 patients with confirmed MS were prospectively included. One radiologist performed manual segmentations of all high-signal intensity lesions using three 3D FLAIR data sets with different TR of 4800 ms (“FLAIR4800”), 8000 ms (“FLAIR8000”) and 10,000 ms (“FLAIR10,000”) and two radiologists double-checked it. The main judgment criterion was the overall number of lesions; secondary objectives were the assessment of lesion location, as well as measuring contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR). A non-parametric Wilcoxon’s test was used to compare the differing FLAIR.
Results
The FLAIR8000 and FLAIR10,000 detected significantly more overall lesions per patient as compared with the FLAIR4800 [116.1 (± 61.7) (p = 0.02) and 115.8 (± 56.3) (p = 0.03) versus 99.2 (± 66.9), respectively]. The FLAIR8000 and FLAIR10,000 detected four and eight times more cortical or juxta-cortical lesions per patient as compared with FLAIR4800 [1.6 (± 2.2) (p = 0.001) and 4.1 (± 5.9) (p = 6 × 10–5) versus 0.4 (± 1.1), respectively]. CNR was significantly correlated to the TR value. It was significantly higher with FLAIR10,000 than it was with FLAIR8000 and FLAIR4800 [16.3 (± 3.5) versus 15 (± 2.4) (p = 0.01) and 12 (± 2.2) (p = 2 × 10–6), respectively]
Conclusion
An optimized 3D FLAIR with a long TR significantly improved both overall lesion detection and CNR in MS patients as compared to a 3D FLAIR with factory settings.