PURPOSE: To demonstrate the clinical and magnetic resonance (MR) imagi
ng findings of brain capillary telangiectasia and compare them with po
stmortem specimens. MATERIALS AND METHODS: MR images obtained in and c
linical histories of 18 adult patients with a presumed diagnosis of ca
pillary telangiectasia examined within 3 years were retrospectively re
viewed. All patients had undergone MR imaging with conventional T1- an
d T2-weighted spin-echo sequences and gadolinium-enhanced T1-weighted
and susceptibility-sensitive gradient-echo (GRE) sequences. No biopsie
s had been performed. Fourteen patients had undergone clinical and MR
imaging follow-up (median, II months). Postmortem tissues from three c
ases of histopathologically confirmed capillary telangiectasia were im
aged, RESULTS: All lesions were small, homogeneously enhancing, and hy
po-to isointense on T1-weighted images and iso-to slightly hyperintens
e on proton-density-and T2-weighted images. None was hypointense on pr
oton-density-or T2-weighted images. All lesions showed marked GRE sign
al loss. None had changed at follow-up. Two patients had multiple clas
sic cerebral cavernous angiomas. The three specimens showed no abnorma
l susceptibility and contained no hemosiderin at tissue analysis. CONC
LUSION: Capillary telangiectasia has mild contrast material enhancemen
t but is otherwise undetectable on conventional MR images. It lacks th
e ''hemosiderin rim'' of cavernous angioma and demonstrates increased
susceptibility only on GRE images, likely owing to blood oxygen-level-
dependent contrast. GRE is essential in diagnosing brain capillary tel
angiectasia, which could otherwise be misdiagnosed as neoplasia, subac
ute infarction, or demyelination.