Application of a Modified Combinational Approach to Brain Tumor Detection in MR Images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Biomedical Engineering

Background:

• Brain tumor detection is a critical challenge in medical research.
• Existing methods for Magnetic Resonance (MR) image analysis have limitations in accuracy and tumor area identification.
• Co-Clustering methods, while useful for block clustering, are not directly suitable for precise brain tumor detection.

Purpose of the Study:

• To introduce a novel algorithm, iterative Co-Clustering and K-Means (ICCK), for enhanced brain tumor detection in MR images.
• To address the limitations of traditional Co-Clustering methods in accurately identifying tumor regions.
• To improve the accuracy and efficiency of brain tumor segmentation using advanced clustering techniques.

Main Methods:

• The proposed algorithm involves image pre-processing and enhancement.
• A modified Co-Clustering method, specifically the latent block model (LBM), is used to identify and isolate tumor-containing image regions.
• The K-Means clustering algorithm is subsequently applied for precise tumor area detection.

Main Results:

• The ICCK algorithm demonstrated high performance metrics on the BraTS2019 dataset.
• Achieved sensitivity of 82.41%, specificity of 99.74%, accuracy of 99.28%, and Dice similarity coefficient of 84.87%.
• The method significantly outperforms existing techniques, particularly on complex medical images.

Conclusions:

• The novel ICCK algorithm offers a superior approach for brain tumor detection in MR imaging.
• The integration of modified Co-Clustering and K-Means effectively overcomes limitations of previous methods.
• This technique shows significant potential for improving diagnostic accuracy in neuro-oncology.