Automatic Brain Tumor Segmentation, and 3D Reconstruction and Visualization Using Augmented Reality
Mohamed Amine Guerroudji, Kahina Amara, Samir Benbelkacem, Adel Oulefki, Nadia Zenati, Djamel Aouam, Oualid Djekoune, Mostefa Masmoudi
Abstract
Augmented reality is the overlay of computer generated images, videos, text, sounds on real world structures. It has previously been used for medical aid diagnosis as a support tool for medical practice/training. Magnetic resonance imaging (MRI) is widely used in medical imaging, especially in the detection of brain tumors. Although it has great advantages, like any other acquisition technique, it suffers from some constraints; such as the problem of noise, which can generate unnecessary and false information, which can escape and deceive the human eye. In this paper, we present an augmented reality (AR) system that simplifies 3D brain tumor reconstruction and visualization. We develop a robust and precise method of segmentation of a large class of tumor in brain MRI images, intending to provide useful information to practitioners, for the diagnosis and therapeutic management. The developed tumor segmentation includes two main components: Pre-processing rather than treatment and segmentation. The pre-processing step involves the operations of reducing the intensity heterogeneity by using image enhancement. The tumor segmentation and detection are provided using the Fuzzy C-Means (FCM) classification algorithm followed by morphological operations. It is based on the hypothesis that the tumor appears with specific gray levels in the image, corresponding to an additional class. The 3D brain tumor reconstruction, using 3D Slicer, was virtually augmented on the real scene. The experimental results using AR visualization show improves significantly In the world of brain tumor.