MRI susceptibility-weighted imaging (SWI) provides better contrast and is typically done at higher spatial resolution than MRI T2*-weighted gradient-recalled echo (GRE), increasing the sensitivity for cerebral microbleed (CMB) detection. However, few studies have implemented SWI and GRE in the same scan session, and there is uncertainty regarding whether findings on SWI are comparable to findings on GRE. We compared the sensitivity and reliability of SWI vs. GRE sequences for detecting CMBs in a community population.
Using a standardized case report form, two radiologists independently identified CMBs on SWI and GRE in 247 participants age 40 to 75 in the community-based Prospective Urban Rural (PURE) MIND substudy. MRI was performed on the same GE 3T MRI scanner using typical clinical sequence parameters. SWI and GRE were read >2 weeks apart. After the independent readings, the two raters met to determine CMB presence by consensus.
Mean age was 58.0±7.6 years, 65% were women, 23% had hypertension and 8.6% had diabetes. By consensus, CMBs were seen in 30 participants (12.2%) on SWI but in only 13 (5.2%) on GRE (p<0.001). Inter-rater reliability was moderately good on SWI (kappa 0.52, 95% CI 0.34-0.71) and GRE (kappa 0.51, 95% CI 0.30-0.73). A pattern of lobar CMBs, with or without cerebellar CMBs but without other non-lobar CMBs, was seen in 8/13 (62%) with GRE CMBs and 19/30 with SWI CMBs (63%). Increased age and diabetes were associated with CMBs on both SWI and GRE.
More than twice as many CMBs were detected on MRI SWI compared to GRE, with the same inter-rater reliability. SWI may be the preferred sequence for detecting CMBs, given its greater sensitivity. However, CMB distribution patterns and risk factors were similar regardless of sequence type, suggesting that the findings of studies using either sequence are comparable.