Authors:

Vedic Sharma

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Traumatic Stress Disorder Post-Traumatic Stress Disorder (PTSD) is a prevalent mental disorder following traumatic incidents. While much research has been done on functional brain changes associated with PTSD, growing data has also indicated that certain structural changes in the brain are associated with this disorder. Advancements in machine learning have provided an effective method to study these structural alterations. The purpose of this project was to train a Convolutional Neural Network (CNN) machine learning model with structural T1-weighted (T1w) MRI scans to predict PTSD Checklist for DSM-5 (PCL-5) scores. Following this, a novel application of explainable artificial intelligence (XAI) techniques was implemented to highlight areas of the brain relevant to the decision making of the model, consequently indicating structural changes in the brain that may be relevant or associated with PTSD. Initial model training with a limited dataset yielded poor model performance in spite of data augmentation techniques. This indicates the need for a larger comprehensive dataset, with upwards of 10,000 images, for better performance. As such, this project is ongoing, with a need to obtain data from external sources including the UK Biobank or Enigma study to acquire the necessary amount of data. 419

Source:

University of Florida / 2024

Topics:

No topics listed

Co-authors:

Vedic Sharma