Pakistan Journal of Engineering & Technology (Sep 2020)
Morphological Segmentation Classification and Extraction of Brain Tumor using Adoptive Water Shed Algorithm
Abstract
The brain tumor is widely seen as cancer which is considered as the second leading cause of human bereavement. Abnormal growth of cells can be found inside or in the boundary of the brain. This kind of abnormality can affect the functionality of the brain or can harm the natural behavior of a person. In general, the brain tumor has two types: one is Benign and another is a Malignant tumor. The Benign tumor cannot spread out suddenly and cannot detriment the other parts of the brain, but the Malignant brain tumor is one type of cancer tumor that can uninterrupted to the patient’s death and it will be prolonged with the worst condition and also affect the neighboring healthy brain tissues. The difficulty is that the tumor cell is not be identified at its initial stage and when they identified it’s difficult to recover the patient in this means patients meet death. This study is contributing to the field of image processing so that the tumor can be identified earlier and with the help of early treatment lives can be saved as the burden on society will be reduced especially in the poor countries. The main technique for tumor identification is MRI imaging. There are many other techniques used for this purpose like CT, MRI, and X-Rays scientists are now working on the new techniques every day a lot of new ideas are developed and implemented. Still, now MRI imaging is a more reliable technique for tumor identification. But there is a need to improve accuracy on which better results are dependent. There is a lot of burden on the doctors to identify and separate the tumor from the images so there is a need to develop an automatic system that will reduce this burden. There is a bundle of algorithms that have been developed to sort this problem. But still now due to some problems or limitations of algorithms it is still unsolved. In this paper, we have proposed an automatic model with a complete framework of tumor identification, segmentation, and classification is proposed with 94% accuracy results achieved by Support Vector Machine (VSM). We used 200 MRI images for this experiment.
