Abstract— Alzheimer’s is considered to be a syndrome where the cognitive functionalities of the human brain decline beyond the expected consequences of ageing. As stated by the World Health Organization (WHO) currently there are nearly about 55 million people who are affected from Dementia. It has been found that not much have been discovered about the consequences of Dementia in India but there are currently nearly 4 million people who are affected with different forms of Dementia in India whereas nearly 44 million people are living with this condition worldwide. In 2019 it became the 9th leading cause of deaths amongst all diseases worldwide leaving behind strokes. Almost 60-70% cases of Dementia are caused by Alzheimer’s alone which makes it one of the leading causes of Dementia. With no cure currently available Alzheimer’s surely needs to be addressed.
Along with the clinical aspects that are related with detection of Alzheimer’s, the psychological and socio- economic impacts are also studied that the disease might have on the affected people with the help of various shallow learning methods in Machine Learning. The dataset used consists of 372 patient’s records. For the deep learning implementation, the dataset comprises of 5121 files belonging to 4 classes. Both the datasets are accessible through Open Access Series of Imaging Studies (OASIS) and also available on Kaggle.
From a combination of linear, distance and tree-based algorithms we have used 12 such algorithms with four clinical features i.e., Mini Mental State Examination (MMSE), Normalized Whole Brain Volume (nWRV), Atlas Scaling Factor (ASF) and Estimated Total Intracranial Volume (eTIV). The analysis have been performed to find certain patterns and correlations in the dataset on the data fields such as age, gender, education social status etc., furthermore an analysis has been performed using CNN, ResNet-50 transfer learning technique to figure out at what stage is the progression of alzheimer is, the stages are namely ‘NonDemented’, 'VeryMildDemented', 'ModerateDemented', and 'MildDemented,
Keywords — Dementia, Alzheimer’s Disease, MMSE, nWRV, eTIV, ASF, Machine Learning, Feature Selection, Deep Learning CNN, ResNet50.
|
DOI:
10.17148/IJARCCE.2022.115153
