AN INTELLIGENT HYBRID APPROACH FOR BRAIN PATHOLOGY DETECTION IN MRI IMAGES

Abstract

Medical Image Processing is a complex and challenging field nowadays. Processing of MRI Images is one of the parts of this
field for efficient brain pathology detection like tumor, asymptomatic unruptured aneurysms, Alzheimer's disease, vascular
dementia, cerebral microbleeds in brain and multiple sclerosis (MS) in magnetic resonance (MR) images. The methodology used
in this paper for brain pathology detection consists of the following steps: The first step includes pre-processing by a Wavelet
Transform (WT) for removal of noises like Salt and Pepper noise, Gaussian, Speckle and Brownian noise, without affecting the
image quality. The second step is to extract the features from the pre-processed image. The process of feature extraction is
carried out by a Walsh- Hadamard Transform (WHT) methodology. The final step involves the detection of abnormality by
segmenting the abnormal tissues using a combined methodology called Modified Fuzzy C-Means Clustering (MFCM) followed
by Level Set (LS). The performance measure of proposed system is evaluated both by objective and subjective method. Feature
extraction and segmentation is evaluated objectively by using confusion matrix and by measuring Accuracy, Sensitivity,
Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), False Positive Rate (FPR) or Over
Segmentation(OvS), False Negative Rate (FNR) or Under Segmentation(UnS), and Total False Rate or Incorrect
Segmentation(InS). Subjective evaluation is done by taking the opinion of 35 expert radiologists that is average mean opinion
score to corroborate the results of proposed method. From the obtained results it is understand that the proposed new amalgam
technique is giving 95% accurate results for detecting abnormality in MRI brain images when compared to other hybrid
methodology.