Probabilistic cortical surface map of the middle cerebral artery territory for single-photon emission computed tomography studies.

OBJECT

The middle cerebral artery (MCA) is the intracranial vessel most frequently affected by cerebrovascular diseases. A more accurate knowledge of the topography of this vessel may have an impact on treatment strategies for ischemic cerebrovascular diseases. The aim of this study was to construct a topographic map of the MCA territory for single-photon emission computed tomography (SPECT) using statistical brain mapping.

METHODS

The margin of the perfusion deficit associated with infarction due to arterial occlusion, as seen on SPECT imaging, is presumed to approximate the borders of the territory of the artery. Basing the study on this hypothesis, SPECT images obtained in 12 patients with large MCA infarctions due to angiographically confirmed MCA trunk occlusion were selected, anatomically standardized, and compared with SPECT images obtained in healthy volunteers to construct probabilistic cortical surface maps of the MCA territory. Crossed cerebellar diaschisis was used as a primary cutoff marker for creation of the map. This MCA map (Method C) was compared with the conventional region of interest (ROI) method (Method A) and previously reported predefined cortical templates (Method B) for preliminary clinical application. The probabilistic cortical surface map of the MCA territory showed that regions with the highest ratio of MCA territory included the transverse temporal gyrus (100%), supramarginal gyrus (100%), and inferior parietal lobule (91-92%). For preliminarily clinical application, this map (Method C) was compared with the conventional ROI method (Method A) in predicting hyperperfusion after carotid endarterectomy by performing a receiver operating characteristic (ROC) analysis, which demonstrated the statistically significant superiority of the MCA map (area under the ROC curve [Az] = 0.91) to the ROI method (Az = 0.75; p = 0.025). The ROC analysis also demonstrated a diagnostic value of the MCA map (Az = 0.95) that equaled predefined cortical templates (Method B) (Az = 0.93).

CONCLUSIONS

The probabilistic cortical surface map of the MCA territory used for SPECT, which was created using statistical brain mapping techniques, would be useful for an objective assessment of the cerebral perfusion status of patients with cerebrovascular diseases.