Alzheimer’s disease diagnosis using convolutional neural networks model

Abstract

The  global  healthcare  system  and  related  fields  are  experiencing  extensive transformations,   taking   inspiration   from   past   trends   to   plan   for   a technologically advanced society. Neurodegenerative diseases are among the illnesses  that  are  hardest  to  treat.  Alzheimer’s  disease  is  one  of  these conditions and is one of the leading causes of dementia. Due to the lack of permanent  treatment  and  the  complexity  of  managing  symptoms  as  theseverity grows, it is crucial to catch Alzheimer’s disease early. The objective of  this  study  was  to  develop  a  convolutional  neural  network  (CNN)-based model to diagnose early-stage Alzheimer’s disease more accurately and with less   data   loss   than   methods previously   discovered.CNN,   is   adept   at processing  and  recognising  images  and  has  been  employed  in  various diagnostic  tools  and  research  in  the  healthcare  sector,  showing  limitless potential.  Convolutional,  pooling  and  fully  linked  layers  are  the  common layers that make up a CNN. In this paper, five CNN modelswere randomly chosen  (ResNet,  DenseNet,  MobileNet,  Inception,  and  Xception)  and  were trained.  ResNet  performed  the  best  and  was  chosen  to  undergo  additional modifications   to   improve   accuracy   to   95.5%. This   was   a   remarkable achievement  that  made  us  hopeful  for  the  performance  of  this  model  in larger datasets as well as other disease detection.