The determination of the embolic source is crucial to understanding the pathogenesis of ischemic stroke, the initiation of appropriate therapy, and the prevention of recurrent infarctions. In this study we undertook to identify the characteristic features on magnetic resonance images of patients who had suffered from stroke due to cardiac embolism (CE), as classified by TOAST (possible and probable). We retrospectively studied magnetic resonance imaging (MRI) findings of patients with ischemic stroke from the Yonsei Stroke Registry (YSR). On the basis of the TOAST classification, 92 patients were identified to have a potential cardiac source of embolism (PCSE), in which 69 patients were found to have high-risk PCSE and 23 patients medium-risk PCSE. To compare their imaging characteristics, another group of 49 patients who were found to have had a stroke due to large artery-to-artery (ATA) embolism-common or internal carotid artery (CCA, ICA)-were identified. Involvement of the simultaneous superficial and deep territories (58.7%; 6.1%, p < 0.001), and combined new anterior and old posterior circulation (15.2%; 2.0%, p = 0.016) were more frequent in PCSE than ATA embolism. Bilateral anterior hemispheric involvement was also more frequent in the PCSE group, but it did not reach statistical significance (13.0%; 4.1%, p = 0.090). ATA embolism tended to involve only superficial territories compared to PCSE (71.4%; 28.3%, p < 0.001). There were no topographic differences between the high-risk and medium-risk groups. With respect to the etiology of PCSE in our population, atrial fibrillation was the most common. Characteristic MRI features of patients with PCSE, which were not documented previously by computed tomography (CT) included: old and new, involvement of multiple different vascular territories, bilateral anterior hemisphere, as well as anterior and posterior circulation. These MRI features, together with simultaneous superficial and deep territorial involvement, help to differentiate the underlying embolic sources, whether they are cardiac or ATA in origin.
All Science Journal Classification (ASJC) codes