To determine the diagnostic accuracy of T1-weighted, FLAIR, and GRE MR imagings in the detection of hemorrhagic transformation in patients with acute cerebral infarction and to compare it with CT. MATERAILS AND METHODS: Fifty-three patients with acute territorial cerebral infarction were studied prospectively. All patients underwent nonenhanced CT and MRI including the T1-weighted, FLAIR, and GRE. Lesion conspicuity of hemorrhage was scored as follows: 0-none; 1-suspicious; 2-sure. CT and MR imagings were reviewed two radiologists respectively. The mean value of the lesion conspicuity in each CT and MR sequences was compared by means of a Wilcoxon signed ranks test. The time intervals between CT and MR imagings ranged from 3 to 14 hours (mean; 7.6 hours).

Hemorrhagic transformation was detected on nonenhanced CT in 26 of 53 patients. In the detection of hemorrhage in patients with acute cerebral infarction, T1-weighted and FLAIR MR imagings were inferior to NECT (p<0.05). By contrast, lesion conspicuity of GRE MR imaging was not different from that of CT (p=0.5). In addition, lesion conspicuity of GRE MR imaging was greater than that of CT in five patients on reader A and two patients on reader B.

GRE MR imaging was superior to T1-weighted and FLAIR MR imagings, equal to nonenhanced CT in the detection of hemorrhagic transformation in patients with acute cerebral infarction.