ABSTRACT
BACKGROUND: Magnetic resonance-guided laser-induced thermal therapy (MRgLITT) is a novel, minimally invasive method currently being used to treat a wide range of intracranial pathologies. No accepted guidelines exist on what the appropriate magnetic resonance imaging (MRI) sequences are for evaluating short-term postablation changes, especially when patients are not able to receive gadolinium. OBJECTIVE: To evaluate which MRI sequences provide the greatest inter-rater reliability and least amount of variability in assessment of ablation volume after MRgLITT for intracranial neoplasms. METHODS: Twenty patients who received MRgLITT were included. Three raters calculated volumetric measurements on postprocedural axial spoiled gradient recalled (SPGR), fluid-attenuated inversion recovery (FLAIR), diffusion-weighted imaging (DWI), and gradient echo (GRE) sequences. Measured volumes were analyzed using intraclass correlation to determine which protocol had the most concordance among the 3 raters. RESULTS: Postcontrast SPGR sequences were most concordant in our study, with an intraclass correlation of 0.981. DWI was the next-most concordant imaging sequence with an intraclass correlation of 0.958. The least concordant were GRE (0.895) and FLAIR (0.866) images. SPGR was also the least variable and had the most consistent volume ratings compared to the other sequences. CONCLUSION: This study is the first to evaluate the inter-rater reliability of different MRI sequence protocols in the context of post-MRgLITT volumetric evaluation. SPGR postcontrast images facilitate the greatest interobserver concordance when characterizing post-MRgLITT tumor appearance and volumetrics, with DWI ranked second. Based on our findings, SPGR sequences are likely to yield the highest degree of concordance in post-MRgLITT lesion evaluation. When gadolinium cannot be given, DWI should provide the next most reliable estimation.
