There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Jaden Wang |
| Institution: | Oakton High School |
| Department: | Science Department |
| Country: | |
| Proposed Analysis: | Millions of people suffer from Alzheimer’s disease, yet still, researchers have not found a cure. Alzheimer's disease is the 6th leading cause of death in the United States. Therefore an early diagnosis plays a vital role in allowing patients to start preventative treatments and provide crucial insight about the disease. The use of deep learning and Alzheimer’s diagnostics is promising as a tool for physicians in making a more accurate and informed diagnosis. A deep learning model with specifically fusion image scans, a combination of two different types of image scans: PET and MRI, could possibly be a better diagnostic tool in identifying between Alzheimer's versus healthy image scans. A fusion of PET and MRI scans can provide a complete picture of anatomical features and brain activity when identifying a Alzheimer's or healthy brain scan. The code will be run and written on Visual Studio Code, for sorting data and the image fusion algorithm, Colab, in order to use their GPU when running the model. The image fusion algorithm that will be used was proposed and published on GitHub by Layez and Susstrunk 2019. The deep learning model will be based on ResNet (residual network) proposed and published Kaming et al. 2016 using specifically ResNet in the Keras package. Other programs such as sorting data and added layers and parameter on the ResNet model specific toward the research would be written individually. The results will be from the model's performance on the test set and derive statistics such as the accuracy, precision, and sensitivity and graphs such as the AUC ROC curve (receiver operating characteristic) and the training and validation accuracy. These statistics can be compared with the results of other similar models using a single image modality. |
| Additional Investigators |
