Diagnosis of intracranial hemorrhagic lesions: comparison between 3D-SWAN (3D T2*-weighted imaging with multi-echo acquisition) and 2D-T2*-weighted imaging

Abstract

3D-susceptibility-weighted angiography (SWAN) can produce high-resolution images that yield excellent susceptibility-weighted contrast at a relatively short acquisition time. To compare SWAN- and 2D-T2*-weighted gradient-echo images (T2*-WI) for their sensitivity in the depiction of cerebral hemorrhagic lesions. We subjected 75 patients with suspected cerebral hemorrhagic lesions to SWAN and T2*-WI at 3T. We first measured the contrast-to-noise ratio (CNR) using an agar phantom that contained different concentrations of superparamagnetic iron oxide (SPIO). The acquisition time for SWAN and T2*-WI was similar (182 vs. 196 s). Neuroradiologists compared the two imaging methods for lesion detectability and conspicuity. The CNR of the phantom was higher on SWAN images. Of the 75 patients, 50 were found to have a total of 278 cerebral hemorrhagic lesions (microbleeds, n = 229 [82.4%]; intracerebral hemorrhage, n = 18 [6.5%]; superficial siderosis, n = 13 [4.7%]; axonal injuries, n = 8 [2.9%]; subarachnoid hemorrhage [SAH] or brain contusion, n = 3 each [1.0%]; subdural hematoma, n = 2 [0.7%]; cavernous hemangioma or dural arterteriovenous fistula, n = 1 each [0.4%]). In none of the lesions was the SWAN sequence inferior to T2*-WI with respect to lesion detectability and conspicuity. In fact, SWAN yielded better lesion conspicuity in patients with superficial siderosis and SAH: it detected significantly more lesions than T2*-WI (P < 0.01) and it was particularly useful for the detection of microbleeds and lesions near the skull base. SWAN is equal or superior to standard T2*-WI for the diagnosis of various cerebral hemorrhagic lesions. Because its acquisition time is reasonable it may replace T2*-WI.