An Automatic Protocol to Detect the Fed & Fasted Brain using Multivariate Analysis of Diffusion Weighted Data Sets

Ania Bentez^1,2, Blanca Lizarbe¹, Luis Lago-Fernndez², Pilar Lpez-Larrubia¹, Sebastian Cerdn¹, Manuel Snchez-Montas²

We present a model for the automatic classification of diffusion weighted images from C57 mouse brain between fed and fasted states. The method requires no preprocessing steps and provides 100% correct classifications of the eight mice of the study between the corresponding fed and fasted classes. The absence of pre-processing steps avoids the possibility to introduce information previously not present in the original image and favors an automatic unbiased interpretation. The approach outlined here may be useful in the investigation of the cerebral causes of obesity and its treatments and could extend to an automated diagnostic imaging system for food intake disorders, obesity, anorexia or bulimia.