A Simple Linear Discrimination Algorithm for AD Patients and Normal Controls
2012 ◽
Vol 155-156
◽
pp. 18-22
Keyword(s):
The Mean
◽
One simple but effective discrimination method was presented in this paper to separate AD from normal controls. After detecting the thickness of cortex with highly significant difference, the mean and standard deviation of these vertices are computed to construct confidence intervals. We introduced one relax coefficients to control the width of intervals and by experiments the coefficients was optimized. Experiments results showed that using this simple method, the classification accuracy, sensitivity and specificity of Alzheimer’s disease versus normal controls could be as high as 85%, 88.89% and 93.84% respectively.