Abstract: In this paper, presents an effective strategy for brain tumor characterization, where, the brain tumor pictures are arranged into typical normal, non-dangerous (Benign) brain tumor and destructive (Malignant) brain tumor. This paper introduces an efficient method approached of brain tumor classification and segmentation, where, the brain tumor images are generally classified into a normal or non-cancerous (benign) brain tumor detection and cancerous (malignant) brain tumor detection. The proposed method follows three steps, (1) pre-processing for Gaussian filter, (2) textural feature extraction for glcm and (3) SOM classification. Gaussian filter is first utilized utilizing for evacuate commotion the brain picture into various levels of rough and itemized coefficients and after that the dim level co-event matrix is framed, from which the surface measurements, for example, vitality, differentiate, relationship, homogeneity and entropy are achieved. The results of co-occurrence matrices are then fed into a SOM (self-organizing map) for further classification and tumor detection with fuzzy partition matrix clustering and segmentation.
Keywords: GUI, MRI, SOM Technique, Fuzzy logic, Confusion Matrix, Benign, Malignant
|
DOI:
10.17148/IJARCCE.2018.769
