Contributions
Abstract: P499
Type: Poster
Abstract Category: Pathology and pathogenesis of MS - Imaging
Background: Magnetic resonance imaging (MRI) with gadolinium is considered a “gold standard” diagnostic test in relapsing-remitting multiple sclerosis (RRMS); however, certain contraindications exist. This study is based on our personal observation that in RRMS patients, demyelinating lesions of outstanding brightness (“ultrabright”) on non-contrast axial Fluid Attenuation Inversion Recovery (FLAIR) MRI are more likely to enhance with gadolinium. Here, we attempted to qualitatively and quantitatively identify the “ultrabright” lesions using axial FLAIR and explore their association with gadolinium enhancement.
Methods: The MRI scans of patients with RRMS from 2010-2015 were pre-reviewed for presence of potentially “ultrabright” lesions. The scan data was extracted as anonymized DICOM images (FLAIR, pre- and post-gadolinium T1). Two radiologists independently identified “ultrabright” lesions on FLAIR sequences. The signal-to-noise ratio (SNR) of each “ultrabright” lesion was calculated as follows: SNR = Mean lesion signal intensity / Air signal standard deviation. One of the radiologists also reviewed the T1 pre- and post-gadolinium images to identify the enhancing lesions (study subject identification numbers were recoded).
Results: A total of 107 lesions was included in the analysis. Seventy seven lesions were identified by both radiologists as “ultrabright” (72%). Kappa statistic of inter-rater reliability was not significant. The lesions that were identified by both radiologists as “ultrabright” demonstrated a strong association with gadolinium enhancement (χ2(1) = 8.863, p = .003). For quantitative method, the odds of the lesion being enhanced increased by 1.02 with every 1.0 increment in signal-to-noise ratio (OR 1.02, 95% CI [1.01-1.03], p < .001). Positive predictive value of “ultrabright” lesions of visually identified by both radiologists was 88%, 95% CI (78-94%); positive predictive value of 100% was achieved using cutoff signal-to-noise ratio of 345.
Conclusions: Both qualitatively and quantitatively identified “ultrabright” FLAIR lesions are associated with gadolinium enhancement. Quantitative measurement of lesions SNR provides unequivocal cutoff value for “ultrabright” lesions and acts as a more reliable predictor of gadolinium enhancement with lower numbers of false positives compared to visual identification.
Disclosure: The authors declare no conflict of interests. The study has not been funded.
Abstract: P499
Type: Poster
Abstract Category: Pathology and pathogenesis of MS - Imaging
Background: Magnetic resonance imaging (MRI) with gadolinium is considered a “gold standard” diagnostic test in relapsing-remitting multiple sclerosis (RRMS); however, certain contraindications exist. This study is based on our personal observation that in RRMS patients, demyelinating lesions of outstanding brightness (“ultrabright”) on non-contrast axial Fluid Attenuation Inversion Recovery (FLAIR) MRI are more likely to enhance with gadolinium. Here, we attempted to qualitatively and quantitatively identify the “ultrabright” lesions using axial FLAIR and explore their association with gadolinium enhancement.
Methods: The MRI scans of patients with RRMS from 2010-2015 were pre-reviewed for presence of potentially “ultrabright” lesions. The scan data was extracted as anonymized DICOM images (FLAIR, pre- and post-gadolinium T1). Two radiologists independently identified “ultrabright” lesions on FLAIR sequences. The signal-to-noise ratio (SNR) of each “ultrabright” lesion was calculated as follows: SNR = Mean lesion signal intensity / Air signal standard deviation. One of the radiologists also reviewed the T1 pre- and post-gadolinium images to identify the enhancing lesions (study subject identification numbers were recoded).
Results: A total of 107 lesions was included in the analysis. Seventy seven lesions were identified by both radiologists as “ultrabright” (72%). Kappa statistic of inter-rater reliability was not significant. The lesions that were identified by both radiologists as “ultrabright” demonstrated a strong association with gadolinium enhancement (χ2(1) = 8.863, p = .003). For quantitative method, the odds of the lesion being enhanced increased by 1.02 with every 1.0 increment in signal-to-noise ratio (OR 1.02, 95% CI [1.01-1.03], p < .001). Positive predictive value of “ultrabright” lesions of visually identified by both radiologists was 88%, 95% CI (78-94%); positive predictive value of 100% was achieved using cutoff signal-to-noise ratio of 345.
Conclusions: Both qualitatively and quantitatively identified “ultrabright” FLAIR lesions are associated with gadolinium enhancement. Quantitative measurement of lesions SNR provides unequivocal cutoff value for “ultrabright” lesions and acts as a more reliable predictor of gadolinium enhancement with lower numbers of false positives compared to visual identification.
Disclosure: The authors declare no conflict of interests. The study has not been funded.