T1-weighted subtraction maps improve the detection of contrast-enhancing lesions in multiple sclerosis

Beyrle, Mirjam ; Sepp, Dominik; Eichinger, Paul; Mühlau, Mark; Berthele, Achim; Engl, Christina; Hemmer, Bernhard; Zimmer, Claus; Wiestler, Benedikt; Kirschke, Jan S. und Kertels, Olivia (2026): T1-weighted subtraction maps improve the detection of contrast-enhancing lesions in multiple sclerosis. In: European Journal of Radiology, Bd. 195, 112576

Abstract

In multiple sclerosis (MS), detection of contrast-enhancing lesions (T1c + L) by magnetic resonance imaging (MRI) provides valuable information for initial diagnosis and supports monitoring of acute disease activity. We investigated the added value of subtraction maps from T1-weighted images (T1_sub) for correctly detecting T1c + L.

Methods

Magnetic resonance images of 555 scans in 286 MS patients were normalized into [0;1] range and co-registered for T1_sub calculation. Ground truth (GT) assessment of T1c + L was established through consensus readings using all available sequences, particularly T1_sub images. Detection accuracy was compared to GT by three independent neuroradiologists, one with and two without available T1_sub.

Results

GT identified 1109 T1c + L, of which 1037 were detected with subtraction maps. Without subtraction maps, approximately half of these lesions were missed. Lesions overlooked without T1_sub were smaller, less hyperintense in post-contrast scans, and more frequently located infratentorial. Without T1_sub, about one contrast-enhancing lesion remained undetected per scan; however, including T1_sub reduced the relative risk of missing a lesion by nearly 80 %. Furthermore, almost a third of examinations were falsely interpreted as free of any T1c + L. In contrast, by T1_sub implementation, the proportion of inconspicuous reports was reduced by nearly 90 %.

Conclusion

In a clinically meaningful way, subtraction maps improve detection of contrast-enhancing MS lesions, particularly when smaller, less intense, and infratentorial.

• Dokumententyp: Zeitschriftenartikel
• EU Funded Grant Agreement Number: 101137154
• Keywords: Multiple sclerosis; Magnetic resonance imaging; Dissemination in time; Image subtraction
• Fakultät: Medizin > Munich Cluster for Systems Neurology (SyNergy)
• Themengebiete: 600 Technik, Medizin, angewandte Wissenschaften > 610 Medizin und Gesundheit
• ISSN: 0720-048X
• Sprache: Englisch
• DFG: Gefördert durch die Deutsche Forschungsgemeinschaft (DFG) - 390857198