Context-Driven Hybrid Frequency Suppression And Pyramid-Gated Segmentation for Enhanced MRI Brain Tumor Analysis

Authors

  • Mrs.C.K. Sangeetha Author
  • Dr.M. Jaikumar Author

Keywords:

Brain TumorSegmentation, CHFS-Net, Deep Learning, Denoising, MRI Imaging, RMPGS-Net

Abstract

Brain tumor diagnosis is a useful technique in early diagnosis and clinical decision-making, as it applies MRI scans, which can be used to detect and analyze abnormal growths of brain tissues. The overall aim of this research is to come up with very precise tumor detection and segmentation in the presence of low contrast, background noise, and irregular shapes of the tumor which in many cases make interpretation of clinical images difficult. The suggested dual network model is based on using Context-Driven Hybrid Frequency Suppression Network (CHFS-Net) to denoise MRI images and Recursive Multiscale Pyramid-Gated Segmentation Network (RMPGS-Net) to segment perfectly. CHFS-Net integrates spatial context learning using transformers and frequency filtering using the wavelet domain to successfully eliminate noise but preserve the structure, as opposed to RMPGS-Net which refines the multiscale features to increase the precision of tumor boundaries. This is a hybrid method that offers good fidelity and high localization of tumor targets. As verified experimentally, the framework has a denoising rate of 93.5% and segmentation rate of 91.9% with a higher clarity and stability than traditional RSAD and Otsu-based techniques. CHFS-Net-RMPGS-Net integration provides solid tumor localization, maintains edges better, and visualize better, and thus would be a useful solution to clinical diagnosis in real-time and in-depth neuroimaging in medicine.

Downloads

Download data is not yet available.

Downloads

Published

2026-09-22

How to Cite

Context-Driven Hybrid Frequency Suppression And Pyramid-Gated Segmentation for Enhanced MRI Brain Tumor Analysis. (2026). Minimax Theory and Its Applications, 11(2s), 734-753. https://journalmta.com/index.php/jmta/article/view/450