An Efficient Meningioma Tumor Segmentation System in WMSN Using UNET-RCNN Classification Approaches

Abstract

Meningioma is the primary type tumors which spreads in different regions of human brain and can be screened using Magnetic Resonance Imaging (MRI) modality method. This paper proposes an efficient framework for the detection and diagnosis of tumor regions in Meningioma brain MRI images. The proposed work is structured into two phases as tumor detection phase and tumor diagnosis phase. The tumor detection phase consists of UNet-Convolutional Neural Networks (CNN) architecture, which performs both classification and tumor region segmentation process. The features are computed from the tumor region segmented image and the computed features are used for the diagnosis of segmented tumor regions into either Early or Severe. The proposed Meningioma tumor detection methods are tested and validated on two different brain image datasets NU and Kaggle in this paper. The segmented tumor regions in meningioma brain images are transferred to the remote unit through the Wireless Multimedia Sensor Networks (WMSN).