| dc.description.abstract | Alzheimer’s Disease (AD) presents a major global challenge, prompting the need
for advanced diagnostic methods. This review evaluates recent deep learning approaches
for AD detection using neuroimaging, focusing on literature from 2018 onward. It assesses
the performance of various deep learning models, including convolutional neural networks
(CNNs), and long short-term memory (LSTM) networks, in detecting AD.
The review highlights both the promise and limitations of these methods, such as
the challenges of limited datasets and training techniques. It underscores the importance of
high-quality datasets and suggests future research should focus on developing benchmark
platforms for model comparison.
An ensemble-based classification model featuring advanced preprocessing and pretrained networks is introduced, achieving 99.75% accuracy on the 2D-ADNI dataset and
96.20% on a separate test set. Domain adaptation on the 2D-ADNI dataset reached highest
99.95% training accuracy, while performance on the 3D-ADNI dataset was 88.8%, and
85.8% with point cloud data.
In summary, this review highlights the effectiveness of deep learning models like
integrating CNNs, 3D CNNs, bidirectional-LSTMs combined CNN, used of ADNI and
point cloud data to in AD detection. It calls for further research to address overfitting and
to optimize model architectures. The review identifies both the promise and limitations of
these methods, noting challenges such as limited datasets and issues with training
techniques. It emphasizes the importance of high-quality data and suggests future research
should address overfitting, optimize model architectures, and explore ways to improve
detection capabilities. Additionally, the development of real-time MRI analysis
applications is recommended, with careful attention to ethical standards and data privacy. | en_US |
