Clinical Performance of Quantitative Susceptibility Mapping in Cerebral Microbleed Detection Relative to 2D GRE.

PURPOSE

To evaluate the potential overestimation of cerebral microbleed (CMB) burden by Quantitative Susceptibility Mapping (QSM) compared to 2D gradient recalled echo (2D GRE), as well as the impact of increased motion degradation due to longer scan times, reduced CMB detection from skull-stripping failures, and the relative visibility of CMBs between techniques.

METHODS

Seventy-nine adult subjects with intracranial hemorrhage underwent same-session brain MRI including 2D GRE and multi-echo GRE for QSM processing, as part of routine clinical care. Images were reviewed by a neuroradiologist and trained research assistant for CMB detection, visibility rating, and anatomical distribution. Motion artifacts and areas of non-visualized brain due to skull-stripping were assessed. Statistical analysis included Wilcoxon signed-rank tests for CMB counts, Mann-Whitney U test for motion assessment, and Fisher's exact testing for anatomical distribution patterns.

RESULTS

QSM showed no significant difference in median CMB counts compared to 2D GRE (1 vs 2, p = 0.175) with strong correlation (r = 0.879, p < 1.65e-26). No significant difference in motion degradation was found between techniques (p = 0.7465). Skull-stripping failures affected only 2% of candidate CMBs, in 5 of 79 (6%) subjects. QSM-detected CMBs showed superior conspicuity (73 vs 33 better visualized lesions, p = 0.00975) with 261 rated equally visible. QSM identified 26 calcifications in 20 subjects, 25 of which were misclassified as CMBs on 2D GRE.

CONCLUSION

QSM demonstrates comparable or slightly lower CMB counts than 2D GRE while offering superior lesion conspicuity and ability to distinguish calcifications, supporting its potential clinical implementation for CMB detection.