Evaluation of MRI Enhancement and Segmentation Techniques for Brain Tumor Detection

The most effective technique for locating and detecting brain tumours is unquestionably magnetic resonance imaging, or MRI. Nevertheless, it should be noted that MRI images, that are in their raw form, usually have noise, low contrast, and invisible boundaries of most parts which make the part of tumor classification and segmentation be less accurate. This research explores the quality improvement of MRI image as well as segmentation accuracy with the use of several image enhancement and segmentation methods that were carried out on BraTS 2021 dataset for the purpose of enhancing brain MRI images and improving the accuracy of tumor segmentation. Methods of transformation like the adjustment of the contrast, reduction of the noise, and unsharp masking were carried out and evaluated by means of the Structural Similarity Index Measure (SSIM). In this list, unsharp masking generated the highest SSIM result, 0.9931, which showed its excellent ability to the parts and structure of the fMRI image. In addition, various segmentation methods such as Active Contour, K-Means Clustering, Canny Edge Detection and hybrid approaches were also carried out. Adaptative Resolution technique turned out to have the best SSIM results, with values as high as 0.98, which means that it is very good at precisely marking the boundaries of the tumor. The findings of this study are also relevant for determining the best preprocessing methods to enhance Deep Learning (DL) models' performance on tasks involving the categorisation of brain tumours.

• Publication year

  2025

• Venue

  2025 World Conference on Cutting-Edge Science and Technology (WCCEST)

• Publication date

  2025-09-24

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1109/WCCEST66994.2025.11389950
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Enhancing Medical Image Denoising: A Hybrid Approach Incorporating Adaptive Kalman Filter and Non-Local Means with Latin Square Optimization

  2024cited by this paper
• Molecularly imprinted polymer composites in biological analysis

  2021cited by this paper
• Magnetic Resonance Imaging Contraindications

  2020cited by this paper
• Utilising Low Complexity CNNs to Lift Non-Local Redundancies in Video Coding

  2019cited by this paper
• A survey of the recent architectures of deep convolutional neural networks

  2019cited by this paper
• Recognition and evaluation of bridge cracks with modified active contour model and greedy search-based support vector machine

  2017cited by this paper
• Brain Tumor Segmentation Using Convolutional Neural Networks in MRI Images

  2016cited by this paper
• Semi-Supervised Classification with Graph Convolutional Networks

  2016cited by this paper
• Deep Learning for Identifying Metastatic Breast Cancer

  2016cited by this paper
• The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)

  2015influential reference
• SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation

  2015cited by this paper
• U-Net: Convolutional Networks for Biomedical Image Segmentation

  2015cited by this paper
• Fuzzy c-means clustering with weighted image patch for image segmentation

  2012cited by this paper
• A non-local algorithm for image denoising

  2005cited by this paper
• Snakes: Active contour models

  2004cited by this paper
• Fundamental enhancement techniques

  2000cited by this paper
• A Computational Approach to Edge Detection

  1986cited by this paper

• No citing papers are available for this paper.