Deep Residual Nets for Improved Alzheimer's Diagnosis
read more
Citations
Convolutional neural networks for classification of Alzheimer's disease: Overview and reproducible evaluation.
Deep learning to detect Alzheimer's disease from neuroimaging: A systematic literature review.
Deep Learning and Neurology: A Systematic Review.
Binary Classification of Alzheimer’s Disease Using sMRI Imaging Modality and Deep Learning
Transfer Learning for Alzheimer's Disease Detection on MRI Images
References
Deep Residual Learning for Image Recognition
Deep Residual Learning for Image Recognition
GradientBased Learning Applied to Document Recognition
Deep Learning-Based Feature Representation for AD/MCI Classification
Predicting Alzheimer's disease: a neuroimaging study with 3D convolutional neural networks
Related Papers (5)
Frequently Asked Questions (13)
Q2. What are the future works in "Deep residual nets for improved alzheimer’s diagnosis" ?
As part of future work, the authors hope to understand more closely the contribution of pretraining versus depth.
Q3. What is the key method for regularizing networks and simulating more data?
a key method for regularizing networks and simulating more data is the use of real-time data augmentation (affine transformations of the data through rotations, flips and translations during training).
Q4. What is the architecture of the network?
The layer-wise architecture enables the network to learn increasingly abstract spatial features to differentiate object categories.
Q5. What is the main shortcoming of existing CNN-based approaches?
The main shortcoming of existing CNN-based approaches is that the networks are shallow, with only a single layer of convolutions to learn a latent feature representation of the data.
Q6. What was the purpose of this study?
Using an 80/20 train/test set split, the authors aimed to assess their hypothesis that pretrained residual networks would improve AD diagnosis.
Q7. How many layers are used to predict the outputs of the network?
The authors add two fully-connected layers, with 1000 and 100 hidden units respectively, that predict three outputs using a softmax classifier.
Q8. What is the architecture for the residual network?
The architecture allows multiple pathways for gradients to flow through the network, which permits the creation of much deeper networks without the burden of vanishing gradients.
Q9. What is the main limitation of the CNN?
This limitation prevents the learning of hierarchical representations of the data, which is crucial to medical tasks where morphological changes are often subtle and multifaceted.
Q10. What is the way to train a neural network?
Automating this analysis for decision support is itself hindered by the limited size of image data sets to help train models, particularly deep convolutional neural networks(CNNs) [3] which have shown promise in many imaging problems.
Q11. What is the way to train a network?
For Q1, the authors trained two networks – the proposed approach in Sec. 2 (pretrained ResNet + augmentation) and a baseline CNN of one convolutional layer containing 5x5 kernels and 64 feature maps, and two fully-connected layers containing 1000 and 100 hidden units, respectively, each with dropout prior to the non-linearity (to approximate existing approaches [1, 4]).
Q12. What is the purpose of this paper?
The authors validate their hypothesis that pretrained deep residual networks improve AD diagnosis by performing 3-way classification (AD vs. MCI vs. healthy) on brain MRIs provided by the Alzheimer’s Disease Neuroimaging Inititative (ADNI).
Q13. What is the name of the paper?
Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.ucla.edu).