CNN-SVM for prediction Alzheimer disease in early step

Alzheimer's Disease (AD) is a progressive neurodegenerative disorder that first affects memory functions and then progressively affects all cognitive functions with behavioral deficiencies and possibly causing death. As the number of patients has increased, its early diagnosis is an objective to be achieved. Nowadays, deep learning, and more particularly transfer learning, is now achieving spectacular performance, unhoped for a few years ago, in areas such as computer vision and the processing of the medical image. In this context, we propose a new study for the early prediction of Alzheimer's disease can be performed using the Magnetic Resonance Imaging (MRI) image and Convolutional Neural Network (CNN). Our contribution was to merge, transfer learning and Support Vector Machine (SVM) technologies to predict Alzheimer's disease at an early stage; the Mild Cognitive Impairment (MCI). We test our method with the collection of OASIS database brain specialized to analyze Alzheimer MRI with 420 subjects; 210 normal and 210 MCI. Our system gave us 94.44% accuracy.