S Mirafzal1, A Goujon1, R Deschamps2, K Zuber3, J C Sadik1, O Gout2, Augustin Lecler4, J Savatovsky1. 1. Department of Neuroradiology, Foundation Adolphe de Rothschild Hospital, Paris, France. 2. Department of Neurology, Foundation Adolphe de Rothschild Hospital, Paris, France. 3. Department of Biostatistics, Foundation Adolphe de Rothschild Hospital, Paris, France. 4. Department of Neuroradiology, Foundation Adolphe de Rothschild Hospital, Paris, France. alecler@for.paris.
Abstract
BACKGROUND: Spinal imaging in multiple sclerosis remains challenging because of its small size and numerous artifacts. OBJECTIVE: To compare 3D Phase-Sensitive Inversion Recovery (PSIR) to a conventional dataset of 3D Short Tau Inversion Recovery (STIR) and T2-weighted imaging at 3 Tesla to detect multiple sclerosis spinal cord lesions. METHODS: This prospective single-center study was approved by a national research ethics board and included 54 patients (median age 44) enrolled from December 2016 to August 2018. Two neuroradiologists individually analyzed the two datasets separately and in random order. Discrepancies were resolved by consensus with a third neuroradiologist. The primary judgment criterion was the number of spinal cord lesions. Secondary judgment criteria included location of the lesions, reader-reported confidence and conspicuity assessed with the lesion-to-cord contrast ratio (LCCR). RESULTS: 3D PSIR detected significantly more lesions than the conventional dataset (371 versus 173, respectively, p < 0.05). Seven patients had no detected lesion with the conventional dataset, whereas 3D PSIR detected at least one lesion. LCCR mean reader-reported confidence (p < 0.001) and inter-observer agreement were higher using 3D PSIR. CONCLUSIONS: 3D PSIR significantly improved overall spinal cord lesion detection in MS patients, with higher reader-reported confidence, higher lesion contrast, and higher inter-reader agreement.
BACKGROUND: Spinal imaging in multiple sclerosis remains challenging because of its small size and numerous artifacts. OBJECTIVE: To compare 3D Phase-Sensitive Inversion Recovery (PSIR) to a conventional dataset of 3D Short Tau Inversion Recovery (STIR) and T2-weighted imaging at 3 Tesla to detect multiple sclerosis spinal cord lesions. METHODS: This prospective single-center study was approved by a national research ethics board and included 54 patients (median age 44) enrolled from December 2016 to August 2018. Two neuroradiologists individually analyzed the two datasets separately and in random order. Discrepancies were resolved by consensus with a third neuroradiologist. The primary judgment criterion was the number of spinal cord lesions. Secondary judgment criteria included location of the lesions, reader-reported confidence and conspicuity assessed with the lesion-to-cord contrast ratio (LCCR). RESULTS: 3D PSIR detected significantly more lesions than the conventional dataset (371 versus 173, respectively, p < 0.05). Seven patients had no detected lesion with the conventional dataset, whereas 3D PSIR detected at least one lesion. LCCR mean reader-reported confidence (p < 0.001) and inter-observer agreement were higher using 3D PSIR. CONCLUSIONS: 3D PSIR significantly improved overall spinal cord lesion detection in MSpatients, with higher reader-reported confidence, higher lesion contrast, and higher inter-reader agreement.