PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
Machine learning algorithms require a large and diverse data set for robust training. However, gathering a sufficient number is a difficult task due to time and budget constraints. Generated data sets can augment training data and provide diverse example for training. We propose a method to generate realistic diffuse optical tomography (DOT) data sets based on known physiological components of the DOT signal. We generate three dimensional models of each signal component and seed the hemodynamic response to activate targeted cortices. Our method reduces the need for a large recruitment process and increases the accuracy of machine learning algorithms.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
The alert did not successfully save. Please try again later.
Zephaniah Phillips V, Seung-Hyun Lee, Eunjeong Choi, Dong Cheon Kim, Ah Song Jang, Hee Kyong Kim, Beop-Min Kim, "Generating whole brain diffuse optical tomography data sets for machine learning data augmentation," Proc. SPIE 12364, Clinical and Translational Neurophotonics 2023, 123640C (17 March 2023); https://doi.org/10.1117/12.2648210