Alzheimer's Disease Staging Using Enhanced Inception-ResNet-V2 and Improved XceptionNet Models for 3D MRI Classification and Segmentation.
Authors
Affiliations (5)
Affiliations (5)
- SCOPE, School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh 522237, India. Electronic address: [email protected].
- Department of Computer Science and Engineering, DVR & Dr HS MIC College of Technology, Kanchikacherla, NTR District, Andhra Prasad.
- Department of Computer Science and Engineering, Amrita Vishwa Vidyapeetham, Amaravati Campus, Guntur, Andhra Pradesh, India.
- School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, India.
- Koneru Lakshmaiah Education Foundation, Greenfields, Vaddeswaram, Guntur District, Andhra Pradesh, Pin: 522502, India.
Abstract
Neurologists have a significant challenge due to the progressive nature of Alzheimer's disease (AD) and its severe effects on cognitive function. Recent advances in neuroimage analysis have opened the door to novel machine-learning techniques that could greatly improve AD diagnosis, progression prediction, and detection. In this research, we provide an enhanced hybrid deep learning approach for combined AD classification and segmentation. An enhanced inception-ResNet-V2 model is used for the multi-class classification of AD and an improved XceptionNet model is used to segment affected brain region of AD. The spatial features present in 3D MRI scans are effectively extracted by the parallel convolutional neural network (PCNN) model. The OASIS and ADNI datasets were used in this research to detect and classify the AD stage. The proposed approach yielded consistently excellent testing accuracy and outstanding training accuracy. For testing, a higher accuracy of 99.5% for the OASIS dataset and 99.7% for the ADNI dataset is attained using the proposed approach. Based on 3D MRI brain scans, these results demonstrate the exceptional ability of the proposed models, especially the Improved XceptionNet, to identify AD reliably. Based on the findings of the experiment, the proposed model outperforms cutting-edge deep learning models for classification and segmentation. The experimental results show that incorporating advanced architectures significantly improves the precision of detecting and assessing brain changes associated with AD, offering practical tools for early diagnosis and monitoring the disease naturally.