Comparison of pre-processing and feature extraction techniques for Alzheimer detection using MRI images

Alzheimer’s disease is a neurological disorder that affects the brain’s working in the elders. The data generated through MRI images vary significantly in image quality, pixel size, the orientation of the image, dimension, and noises, and several disturbances degrade the data quality for further processing. To minimize the impact of Alzheimer disease, which does not show any symptoms in the early stages, early detection of the disease is the primary requirement. The image dataset needs to be procured and improvised to obtain the maximum information and accurately interpret data. Several pre-processing techniques like image correction, standardization, resizing, thresholding etc. improvise the readability of the images and produces uniformity in the dataset. The application of an accurate feature extraction technique further allows the adequate classification of the dataset and identification of the implication of variation in image characteristics. The critical variation in the sequence of images can be determined using the feature extraction technique. In this study, several pre-processing and feature extraction techniques are investigated thoroughly to identify the most appropriate technique for the enhancement of dataset and model classification accuracy.