A Novel Thresholding Approach to Find Out the Brain Tumor Region from MR Images

Abstract

The goal of this work concentrates on MRI (Magnetic Resonance Imaging) brain tumor findings by utilizing the BRATS dataset images. At first, it undergoes preprocessing criteria using an enhanced DWT (Discrete Wavelet Transform) filtering to decrease the noise level in the picture. Following this, segmentation has to extricate the tumor region dependent on improved thresholding activity. At that point, its effectiveness has processed by using the performance measurements, for example, Normalized cross-correlation, PSNR (Peak Signal to Noise Ratio), NAE (Normalized Absolute Error), and AD (Average Difference). The reproduction results show that unrivaled outcomes on the proposed thresholding task contrasted with comparable techniques. Furthermore, this was better than the other strategies for both Rician and Gaussian noise. The proposed approach supports the experts in identifying the exact location of the tumor and broadening their lifetime.

KEYWORDS:

• Brain tumor
• Enhanced DWT filtering
• MR images
• Performance evaluation measures
• Threshold based segmentation

Additional information

Notes on contributors

R. Remya

R Remya received PhD, ME and BE degree's in ECE and applied electronics in the year 2021, 2012 and 2009, respectively. She has been with the dept of ECE as assistant professor at Arunachala College of Engineering for Women, India. Her field of interest is image processing. She has published four papers.

K. Parimala Geetha

K Parimala Geetha received the PhD, ME and BE in ECE in the year 2010, 1999 and 1990, respectively. She is working as professor in the Department of ECE in Nalla Malla Reddy Engineering College, Hyderabad. She has published 15 papers in national and international conferences. Her area of interest includes image processing, VLSI design, medical image processing, optical communication and optical image processing. Six candidates have been awarded PhD under her guidance. Email: [email protected]