Learning-based improved seeded region growing algorithm for brain tumor identification

This paper presents a novel approach to segment Magnetic Resonance Image (MRI) of Brain and identification of brain tumor from the brain MRI. The presented work is based on a novel approach that identifies brain tumor from brain MRI in two stages: initially a brain MRI is processed from generation of threshold T2 and PD image of a brain MRI using Seeded Region Growing algorithm, finally, both images are processed further to classify into tumor images or normal images by using Markov Logic (ML) algorithm. Classification on the basis of tumor’s existence or not in a brain MRI is performed by subtracting a tumor-affected image from a standard image and the resultant image spots to ascertain the existence of tumor in the image. The proposed approach was tested with various datasets of MRI. Results of this research using the designed approach marked out better than the other existing approaches as results are 96.71% for SNR 5DB and 99.96% for SNR 10DB.