Detecting Alzheimer's Disease in its Early Stages Using Machine Learning

The ASDRP Summer 2022 Semester Research Symposium and Expo showcases some of the original scientific findings of 625+ high school students, who conducted research across 80+ projects in a wide array of STEM fields. Students come from throughout Bay Area, across the U.S., and internationally! The Symposium features students presenting the research they completed throughout the summer during their participation in ASDRP.

Aditya Rao, Sriram Natarajan, Vanessa Yao, Shreeya Garg, and Nandini Dharwadkar present their project "Detecting Alzheimer’s Disease in its Early Stages Using Machine Learning" at the Summer 2022 Virtual Blitz Talks Symposium and Expo.

Abstract:
Alzheimer’s Disease is a form of dementia that causes the slow deterioration of the brain as clumps and tangles of proteins form and neural tissues degenerate. This disease affects specific brain sections including the hippocampus, cerebral cortex, and temporal lobe. Because deterioration is a slow process, most patients are often unaware of their condition until severe symptoms show or even death occurs. These deteriorations are ambiguous in MRI scans, and therefore, cannot be distinctly diagnosed as symptoms of Alzheimer’s Disease. In this study, MRI scans were fed into a machine learning algorithm to build a deep-learning model that can accurately classify demented and non-demented brains. Around six thousand MRI scans were implemented in open-source software and a graphics processing unit, ranging from very mildly demented to moderately demented brains in one category. Because of the high volume of images, the algorithm was run in multiple epochs with batches of thirty-two images at a time to create a model with high accuracy. Using this data, the machine learning algorithm is expected to create interpretable visualizations of the risk of Alzheimer’s Disease as well as provide more accurate diagnoses for even the mildest case of Alzheimer’s Disease.

For all things ASDRP: https://www.asdrp.org
Our YouTube Channel:    / @asdrpresearch