A method is disclosed for identifying potential perfusion defects in a tissue region through which blood flows in an object under investigation, based on at least one high-energy image data set covering the tissue region and at least one low-energy image data set covering the tissue region. A virtual contrast medium image data set is established based on the high-energy image data set and the low-energy image data set. Furthermore, first candidate perfusion regions within the virtual contrast medium image data set, and second candidate perfusion defect regions within a further image data set based on the high-energy image data set and/or the low-energy image data are detected, the first candidate perfusion defect regions being compared with the second candidate perfusion defect regions and, based on the comparison, potential perfusion defects being identified. Further disclosed are a corresponding image analysis apparatus and a computed tomography system.