00054
AUTOMATED DIAGNOSIS OF DEMENTIA THROUGH SUPERVISED LEARNING CLASSIFICATION ALGORITHM
AUTOMATED DIAGNOSIS OF DEMENTIA THROUGH SUPERVISED LEARNING CLASSIFICATION ALGORITHM
Friday, February 17, 2017
Exhibit Hall (Hynes Convention Center)
The prevalence of Alzheimer's disease and other forms of dementia is currently a worldwide threat to the health and financial sectors. Dementia is an overarching category of diseases which is categorized by the commonality of memory loss. Some prominent forms include Alzheimer's disease, Parkinson's disease, and Mild Cognitive Impairment. Worldwide, over 47.5 million people are currently affected by some form of dementia. Additionally, in the United States alone, over 5.3 million people are living with Alzheimer's disease, but approximately 55% of these cases are left undiagnosed. In 2015, care for this disease cost the United States an estimated $226 billion. In order to improve both cost efficiency and reduce the time for formal diagnosis, a computer-aided program which could automatically diagnose the presence of several forms of dementia was developed. Utilizing the image processing capabilities of MATLAB, several statistics from axial magnetic resonance images of the patient's brain were extracted. This information was then combined with various other characteristics such as age, gender, and MMSE scores in order to determine an output diagnosis. The program was initially hypothesized to accurately diagnose cases of dementia 90% of the time. After completion of 1000 trials, the program exceeded the initial hypothesis and accurately diagnosed the presence and type of dementia 96% of the time. This project will be furthered by the creation of a mobile application, alongside implementation into clinical trials and an increased number of types diagnosed. Due to its comprehensive approach of diagnosing multiple forms of dementia, this program currently surpasses any competing products on the market while advancing previous research done in the field.