Cheng-Yu WU, Jenn-Lung Su, Lung Chan and Yao-Hsun Tein
Alzheimer's disease (AD) is caused by the deterioration of the brain nerves. Due to the irreversible of AD, it’s important to diagnose accurately as early as possible. In this work, a computer-aided detection (CAD) system was developed to not only predict the Clinical Dementia Rating(CDR) of patients, but provide physicians with more objective quantitative data from CT images to assess the degree of deterioration of AD.
This CAD system was developed based on the correlation between Clinical Dementia Rating (CDR) and the brain atrophy ratio. Volume of brain parenchyma, specific sulcus, volume ratio of ventricle to skull from CT images were calculated through image pre-processing, region growing, threshold, image enhancement, and morphological skeleton. All of these parameter was then used in Support Vector Machine(SVM) to interpreting CDR scores. 60 sets of CT images were used (40 sets as training groups & 20 sets as test groups) to train and test this system. Effectiveness of the system and classifiers was performed by using the comparison with CDR and MRI images of patients, and receiver operating characteristic(ROC)analysis.
SVM classifiers were trained in 11 groups of parameters such as brain parenchyma ratio, left and right brain parietal sulcus, left and right brain central sulcus, left and right lateral sulcus, cerebral ventricle, third ventricle, cerebral cistern and whole ventricle. 20 groups of testing data were used to test the efficiency of the classifier. The accuracy, sensitivity, specificity, and Kappa value of the obtained classifier were 80%, 86.6%, 84.6%, and 0.547, respectively; if the cases with significant difference in brain parenchymal volume ratio and CDR level were excluded, the efficiency could up to 88%, 100%, 84.6% and 0.699, respectively.
Results show that the system not only has a grate diagnosis performance, but also has similar system performance compared to mainstream MRI-based algorithms.