An approach to differentiate Alzheimer’s conditions using MR image–based zernike moments and fractal features

Alzheimer’s disease (AD) is a chronic neurodegenerative disorder that affects a large population. The early detection of mild cognitive impairment (MCI) can help to diagnose AD and assist in further treatment strategies. The morphological alterations of lateral ventricles and the corpus callosum are considered a significant imaging biomarker for the early diagnosis of MCI and AD. Shape-based features provide distinct morphological variations of brain structures during disease progression. In this research, an attempt has been made to analyze the variations of lateral ventricles and the corpus callosum using Zernike moments and fractal box count features. These shape features help to classify the morphological structure of lateral ventricles and the corpus callosum as control, MCI, and AD. The proposed method can quantify the shape variations using Zernike moments and fractal box count features. Here, Zernike moments and fractal box count measures show differences in mean values for control, MCI, and AD subjects with a statistical significance of p<0.05. Performance is analyzed using multilayer perceptron, K-nearest neighbors (KNN) and linear support vector machine (SVM) classifiers. Linear SVM provides better differentiation for controls vs. MCI and controls vs. AD on the order of 93% and 98%, respectively. Results show that the classification of shape descriptors performs better with respect to accuracy, specificity, and sensitivity measures. Therefore, this study is clinically important for the early diagnosis of AD.