Lightweight CNN Approach for Multiclass Alzheimer Stage Classification from MRI Scans

Alzheimer's disease is a perpetual cerebral ailment that gradually impairs brain function.With the aging of the world's population, this condition is growing, yet no cure has been discovered.Early discovery of the condition and subsequent medication may spare a patient from completely losing their memory.Accurate detection relies on MRI and PET scans.This study proposes a cnn framework consisting of just six layers to facilitate classification.Here, we've used a dataset of 6400 neuronal MRI scans acquired via Kaggle repository to assess our model.In the context of accurateness and the quantity of parameters, the proposed framework has been compared with a few other existing works.We have meticulously compared this model's efficiency and accuracy with other pre-trained models such as VGG16, Alexnet, and other custom CNN models such as ADDNet and DemNet.Notably, despite having fewer parameters like 502,276, our model attains a remarkable accuracy score of 99.51%, on the Kaggle's Alzheimer dataset.