Brain tumor detection in MR image using superpixels, principal component analysis and template based K-means clustering algorithm

In the present era, human brain tumor is the extremist dangerous and devil to the human being that leads to certain death. Furthermore, the brain tumor arises more complexity of patients life with time. As a result, early detection of tumors is most crucial to save and prolong the patient’s lifetime. Therefore, enhanced brain tumor detection is required in medical fields. Automatic human brain tumor detection in magnetic resonance imaging (MRI) is playing a vital role in several symptomatic and cures applications. However, the existing schemes (e.g., random forest, Fuzzy C-means, artificial neural network (ANN) and wavelet transform) can detect brain tumors with insufficient accuracy and longer execution time (in minutes). In this paper, we propose an enhanced brain tumor detection scheme based on the template-based K-means (TK) algorithm with superpixels and principal component analysis (PCA) which efficiently detects the human brain tumors in lower execution time. At first, we extract essential features using both superpixels and PCA which helps accurately to detect brain tumors. Then, image enhancement is done using a filter that helps to improve accuracy. Finally, the image segmentation is performed through TK-means clustering algorithm to detect the brain tumor. The experimental results show that the proposed detection scheme achieves a better accuracy and a reduced execution time (in seconds) than other existing schemes for the detection of brain tumor in MR image.