Authors: Ms.Dhanushni.N, Ms.Vivisha Catherin.P
Abstract: Alzheimer’s disease (AD) is a pressing global issue, It’s known as the severe neuron disease. They Mainly damages the Brain cells, which leads to permanent lose of memory which is also called dementia. Many people die due to this disease every year because it is not curable but the early detection can prevent from spreading. Alzheimer’s are most commonly found in the elder peoples or from the age of (60 and above). It requires an efficient and automated system which can detect the disease and classify it in the basis of Alzheimer’s stages like Mild Demented(MD), Moderate Demented(MOD), Non Demented(ND), Very Mild Demented(VMD). For the prediction we use Machine learning and deep learning Algorithm’s like convolutional neural networks for imaging data(CNNs), Random forest and Gradient Boosting(XGBoost / LightGBM), Support Vector Machines(SVM) Which is much more efficient from the preexisting models of the Alzheimer Detection. Of relying on methods, like CNNs and SVM for our model design like Random Forest and XGBoost do typically with fixed structures and manual feature selection processes; we take a different approach thats more intricate and advanced by utilizing transfer learning through the InceptionV3 network already trained on ImageNet for its robust feature extraction abilities. To boost our models effectiveness in handling datasets adequately; we integrate various data augmentation methods such as adjusting image angles and proportions along, with mirroring techniques. Address the issue of class distribution by adjusting the weights for classes to focus more on identifying cases of Alzheimers disease accurately. In addition, to this adjustment in class weighting strategy consider implementing techniques like dropout regularization method and early stopping along with model checkpoint mechanism to prevent the model from learning noise and improve generalization. This holistic strategy leads to a model that's proficient in reducing both positives and false negatives which is crucial, in accurate medical diagnosis.
