Segmentation of Glioma Tumours Using Deep CNN Architecture

EasyChair Preprint 8397

Abstract

We present a fully automated model for the task of segmentation and classification of Glioma tumours based on Deep CNN architecture and Fractal dimensional analysis. Glioma tumours are heterogeneous in shape and vary in location & early diagnosis of gliomas is essential to improve the treatment procedures. The proposed model is the result of a through examination of shortcoming of existing models for similar applications. The suggested approach uses 3-D MRI scans from the BraTS 2015 dataset to divide the tumour into four regions: edema, enhancing, non-enhancing and necrotic, as well as classify whether it is a High-Grade Glioma or Low- Grade Glioma. This model is computationally efficient and allows its adoption in a variety of research.

Keyphrases: CNN, Glioma, MRI, Tumours

@booklet{EasyChair:8397,
  author    = {V Vinamraa and Th Sandeep and K Nayana},
  title     = {Segmentation of Glioma Tumours Using Deep CNN Architecture},
  howpublished = {EasyChair Preprint 8397},
  year      = {EasyChair, 2022}}