Simultaneous Arterial Spin Labeling Cerebral Blood Flow and Morphological Assessments for Detection of Alzheimer's Disease

Rationale and Objectives: The relative roles of arterial spin-labeling (ASL) perfusion imaging and magnetic resonance morphological assessment in diagnosing Alzheimer's disease (AD) have not been established. Our purposes were to directly compare the diagnostic performance of ASL regional cerebral blood flow (rCBF) measurement and that of morphological assessment, and to determine whether or not the combination of the two methods improves diagnostic performance. Materials and Methods: We analyzed 23 consecutive, retrospectively identified AD patients and 23 healthy control subjects. For each subject, both high-resolution T1-weighted images and ASL perfusion images were obtained. A linear discriminant analysis was performed to distinguish the AD patients from the control subjects based on the three imaging parameters: 1) globally normalized gray matter (GM) density determined by voxel-based morphometry (VBM) procedures, 2) normalized rCBF calculated from ASL data, and 3) the combination of the two. The discriminative abilities of these methods were evaluated by the area under the curve (AUC) derived from receiver-operating characteristics analysis. Results: The morphological assessment based on the globally normalized GM density resulted in an AUC of 0.779, whereas ASL-normalized rCBF analysis achieved better performance (AUC = 0.893). The combination of the two methods performed better (AUC = 0.919) than either method alone. Conclusion: Normalized rCBF measurement by ASL may perform better than morphological analysis based on the VBM procedure in discriminating AD patients from healthy control subjects. The combination of the two approaches was more effective than either method alone.