PINet: Privileged Information Improve the Interpretablity and generalization of structural MRI in Alzheimer's Disease.

The irreversible and progressive atrophy by Alzheimer's Disease resulted in continuous decline in thinking and behavioral skills. To date, CNN classifiers were widely applied to assist the early diagnosis of AD and its associated abnormal structures. However, most existing black-box CNN classifiers relied heavily on the limited MRI scans, and used little domain knowledge from the previous clinical findings. In this study, we proposed a framework, named as , to consider the previous domain knowledge as a , and open the black-box in the prediction process. The input domain knowledge guides the neural network to learn representative features and introduced intepretability for further analysis. used a Transformer-like fusion module to iteratively calculate the correlation of the features between image features and PI features, and project the features into a latent space for classification. The module served as a verification to highlight the significant features on the input images. was suitable for neuro-imaging tasks and we demonstrated its application in Alzheimer's Disease using structural MRI scans from ADNI dataset. During the experiments, we employed the abnormal brain structures such as the Hippocampus as the , trained the model with the data from 1.5T scanners and tested from 3T scanners. The F1-score showed that was more robust in transferring to a new dataset, with approximatedly 2% drop (from 0.9471 to 0.9231), while the baseline CNN methods had a 29% drop (from 0.8679 to 0.6154). The performance of was relied on the selection of the domain knowledge as the . Our best model was trained under the guidance of 12 selected ROIs, major in the structures of and . In summary, PINet considered the domain knowledge as the to train the CNN model, and the selected introduced both interpretability and generalization ability to the black box CNN classifiers.