3-D brain segmentation towards the integration of DTI and MRI modalities

Abstract

This study introduces a 3-D segmentation method together with a graphical user interface (GUI) as means to effectively automate the process of segmentation with the ultimate objective of integrating and visualizing diffusion tensor imaging (DTI) with magnetic resonance imaging (MRI) in a fully automated 3-D brain imaging system. A secondary objective is to reduce significantly the segmentation time required to extract key landmarks of the brain in contrast to the manual process currently used at many hospital settings. The results provided will prove this important assertion. The inter-correlation coefficient revealed 96.1% accuracy in segmenting all of the processed data, which consequently led to effective registration of the DTI and MRI modalities since they involve the same landmarks. The average speed of segmentation was just 35 seconds, a reduction of over 20 times of what is required for manual segmentation. In order to create a highly integrated interface, the segmentation results serve as input to a registration algorithm we are currently investigating and whose preliminary results support the significance of relying on an effective segmentation process. T1-weighted 3D Gradient Echo MR and DT images from 16 patients at Miami Children's Hospital were used for evaluation purposes.