Comparison of SWI with T2* in the Diagnosis of Cerebral Amyloid Angiopathy Using ARIA Scoring and Boston Criteria 2.0 at 7T MRI

BACKGROUND AND PURPOSE: 7T-MRI enables detection of subtle hemorrhagic biomarkers associated with cerebral amyloid angiopathy (CAA). However, the relative performance of T2* compared with susceptibility-weighted imaging (SWI), as well as the applicability of Boston Criteria 2.0 at 7T, remains uncertain. This study evaluated diagnostic agreement between T2* and SWI, interobserver variability, and the utility of the Boston Criteria 2.0 at 7T-MRI and characterized the prevalence of emerging imaging biomarkers, such as the intragyral hemorrhage sign (IGHS). MATERIALS AND METHODS: In this retrospective study, 130 examinations from 111 patients referred for 7T-MRI between 2022 and 2025 for transient focal neurologic deficits, cognitive impairment, suspected CAA, or Alzheimer's disease were analyzed. Two experienced neuroradiologists independently evaluated T2* and SWI sequences using ARIA-based categorical scoring for intracerebral parenchymal hemorrhage (IPH), cortical/subcortical microbleeds (C/SC-MB), deep microbleeds (D-MB), cortical superficial siderosis (cSS), and IGHS. Discrepancies were resolved by consensus. Boston Criteria 2.0 classifications were assigned by a vascular neurologist. Intersequence and interobserver agreement were assessed, and associations between imaging markers and Boston classifications were analyzed. RESULTS: Interobserver agreement of neuroradiologists was high across both sequences, with almost perfect agreement for IPH, C/SC-MB, D-MB, and cSS (κ = .85-1.00). T2* and SWI demonstrated almost perfect agreement for IPH, C/SC-MB, D-MB, and cSS (κ = .90-1.00) and substantial agreement for IGHS (κ = .62). Almost perfect agreement was observed between Boston scores based on SWI and T2* (κ = .92). IGHS was observed in about 24% and 41% cases with a positive Boston score with T2* and SWI, respectively. Among 81 patients with at least one hemorrhagic marker, 35 (45%) were excluded from CAA due to D-MBs detection despite radiologic features suggestive of CAA. CONCLUSIONS: T2* and SWI provide comparable diagnostic performance for CAA-related hemorrhagic markers and almost perfect agreement regarding the Boston Criteria at 7T. Interobserver agreement was high across all sequences, supporting the reproducibility of ARIA-based scoring at 7T. Boston Criteria 2.0 are applicable at 7T; however, improved detection of deep microbleeds may lead to false-negative classifications. Updated diagnostic frameworks tailored to 7T-MRI, including potential incorporation of IGHS, may be warranted.