Brain MRI Image Analysis for Alzheimer’s Disease Diagnosis Using Mask R-CNN

Abstract

Alzheimer's disease is the most common form of dementia and ranks as the fifth greatest cause of death among individuals aged 65 and older. Moreover, based on official government records, there have been a substantial rise in the number of deaths associated with Alzheimer's disease. Thus, identifying Alzheimer's disease at an early stage has the potential to increase the chances of survival for individuals affected by the condition. The utilization of Magnetic Resonance Imaging (MRI) in conjunction with machine learning techniques has enhanced and accelerated the process of diagnosing Alzheimer's disease. Nevertheless, the use of manual feature extraction techniques on MRI images using conventional machine learning methods is difficult and requires the guidance of an experienced user. Thus, a method could automate the procedure and decrease the need for feature extraction by employing deep learning technology as an autonomous detection and extraction mechanism. This research employs the Mask-RCNN, a convolutional neural network method, to showcase its capability in segmenting and recognizing objects in 40 MRI pictures from the train and test datasets, which consist of both significantly demented and non-demented cases. The earliest applications for Mask-RCNN were the detection of objects in MRI images and the categorization of object instances. This experiment showcases the suitability and superiority of Mask R-CNN for the diagnosis of dementia in brain MRI images, achieving an impressive accuracy of 97.46%.