المستخلص: |
this research paper proposes a fully automated intelligent classification technique to identify brain tumors that appears in MR1 brain slices. This technique could help both manual interpretations of tumor through visual examinations made by radiologists /physicians which may lead to several missing diagnosis and mistakes due to human errors; it may also help surgeons, systems for Computer aided surgeries and also may help in creating a fully automated system for brain tumor surgeries. A fully automated intelligent classification system is proposed to identify exactly the abnormal volume of the brain, first the brain slice image passes through several processing algorithms as Pulse Coupled Neural Networks (PCNN) and Fuzzy Image Labeling to generate image set of brain parts, afterwards those parts pass over some advanced classifications techniques as Support Vector Machine (SVM) or Artificial Neural Networks applied (ANN) against those brain parts images to identifying the abnormal part of the brain and highlighting it. In this research also we compare the accuracy of each of the classification techniques SVM and ANN to identify the best classification algorithm to be used.
|