DcGAN and EfficientNetB3 Based Analysis and Detection of Alzheimer Detection Using MRI Images

Anxiety, depression, and cognitive decline are among the physiological changes brought on by ongoing stress in daily life. AD is the most prevalent kind of dementia in the elderly and is typified by a progressive loss of memory, confusion, and diminished language and cognitive abilities. The suggested technique combines an EfficientNetB3 model with a Deep Convolutional GAN (DcGAN) in a multi-stage manner. 1,125 labeled samples covering various stages of AD are included in the dataset, which was obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) on Kaggle. To overcome data shortage issues, synthetic pictures are produced with the DcGAN. The EfficientNetB3 framework is subsequently adjusted for AD classification by integrating these artificial pictures with the actual ADNI dataset. The model is assessed using performance measures including accuracy, recall, F1-score, precision, and validation accuracy. 90% accuracy is shown overall in the testing findings, with noteworthy recall and precision measures for the non-demented group. The validation stage's superior accuracy of 93.70% and the testing phase's excellent accuracy of 90.27% demonstrate the model's outstanding generalization capabilities. The thorough outcomes demonstrate the model's ability to solve problems with data shortage by highlighting its competence in both generating and classification tasks. A viable approach to enhancing AD classification with possible diagnostic applications is the use of synthetic pictures in the classification pipeline. This paper concludes by outlining the prospects for furthering clinical procedures and research on AD in the future.