|
|
||||||||
|
|
|||||||||
|
|
Neurosciences: NeurologyNeurology Posters |
1 Radiology and Nuclear Medicine; 2 Institute for Aging Brain and Cognitive Disorders, Hyogo Brain and Heart Center, Himeji, Hyogo, Japan; 3 Radiology and Bioengineering, University of Washington, Seattle, Washington
942
Objectives: To develop a computer-assisted diagnostic system to distinguish Alzheimer disease (AD), Lewy bodies (DLB), and other degenerative disorders in mild dementia patients.
Methods: IMP-SPECT images were obtained from mild degenerative dementia. First, data sets from 20 mild AD, 15 dementia with Lewy bodies, and 17 normal controls were used to develop an automatic diagnosing system using a statistical brain mapping method. AD and DLB specific regional templates were created using 3D-SSP, and critical Z scores in the templates were established. We then used data sets from 50 AD patients, 8 DLB patients, and 10 non-AD/DLB type degenerative dementia (5 FTD and 5 PSP) to test the diagnostic accuracy of the optimized automated system in comparison to the diagnostic interpretation of 3D-SSP Z score maps. These comparisons were performed for differentiating AD and DLB from non-AD/DLB and for differentiating AD from DLB. A receiver operating characteristic (ROC) analysis was performed.
Results: The area under the ROC curve and accuracy for the automatic diagnosis system were 0.915 and 83.6% for AD/DLB vs. non-AD/DLB, and 0.753 and 68.9% for AD vs. DLB patients. The mean area under the ROC curve and accuracy of the visual inspection was 0.840 and 77.1% for AD/DLB vs. non-AD/DLB and 0.679 and 63.7% for AD vs. DLB patients, respectively.
Conclusions: This system achieved better discreminatios of AD, DLB, and other degenerative disorders in mild dementia patients in comparison to commonly performed visual inspection of Z-score maps. This system would help improve differential diagnosis of mild dementia patients.
| ||||||||||||||||||||||||||||||||||||||