Improving Classification Rate of Schizophrenia Using a Multimodal Multi-Layer Perceptron Model with Structural and Functional MR

Abstract

The wide variety of brain imaging technologies allows us to exploit information inherent to different data modalities. The richness of multimodal datasets may increase predictive power and reveal latent variables that otherwise would have not been found. However, the analysis of multimodal data is often conducted by assuming linear interactions which impact the accuracy of the results.